wassname/detect_bs_text
1
0
Fork 0
mirror of https://github.com/wassname/detect_bs_text.git synced 2026-06-27 18:04:45 +08:00
Code Issues Packages Projects Releases Wiki Activity

longer samples, splitting

Browse Source
This commit is contained in:
wassname
2024-01-04 07:44:53 +08:00
parent 96a5e8b4bc
commit 4f9dab2871
22 changed files with 15558 additions and 3776 deletions
Download Patch File Download Diff File Expand all files Collapse all files
nbs/01_detection_using_adapter_ft copy.ipynb
+7167
View File
File diff suppressed because one or more lines are too long
nbs/01_detection_using_adapter_ft.ipynb
+7334 -3732
View File
File diff suppressed because one or more lines are too long
nbs/02_detection_using_tldr_prompt.ipynb
+44 -42
View File
@@ -12,7 +12,7 @@
},
{
"cell_type": "code",
"execution_count": 2,
"execution_count": 1,
"metadata": {},
"outputs": [
{
@@ -34,7 +34,7 @@
},
{
"cell_type": "code",
"execution_count": 3,
"execution_count": 2,
"metadata": {},
"outputs": [],
"source": [
@@ -54,34 +54,35 @@
},
{
"cell_type": "code",
"execution_count": 4,
"execution_count": 3,
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"[PosixPath('../samples/cicero_fin1.md'), PosixPath('../samples/disney_appointment.md'), PosixPath('../samples/fake_paper.md'), PosixPath('../samples/fauci_emails.md'), PosixPath('../samples/harvard_announcement_reminders.md'), PosixPath('../samples/how_to_catch_a_liar.md'), PosixPath('../samples/lk-99_end.md'), PosixPath('../samples/lk-99_espanol.md'), PosixPath('../samples/lorem_ipsum.md'), PosixPath('../samples/openai_board_ann.md'), PosixPath('../samples/openai_paper_weak_to_strong.md'), PosixPath('../samples/politics_is_the_mind_killer.md'), PosixPath('../samples/statement_vyKamala_on_passing_of_johnson.md'), PosixPath('../samples/survey_of_rumours.md')]\n"
]
},
{
"data": {
"text/plain": [
"{'name': 'bad_ml',\n",
" 'url': 'https://arxiv.org/abs/2312.10868',\n",
" 'text': 'This roadmap survey has embarked on an exploration of the\\ntransformative trends in generative AI research, particularly focusing on speculated advancements like Q* and the progressive strides towards AGI. Our analysis highlights a crucial paradigm shift, driven by innovations such as MoE, multi-modal learning, and the pursuit of AGI. These advancements signal a future where AI systems could significantly extend their capabilities in reasoning, contextual understanding, and creative problem-solving. This study reflects on AI’s dual potential to either contribute to or impede global equity and justice. The equitable distribution of AI benefits and its role in decision-making processes raise crucial questions about fairness and inclusivity. It is imperative to thoughtfully integrate AI into societal structures to enhance justice and reduce disparities. Despite these advancements, several open questions and research gaps remain. These include ensuring the ethical alignment of advanced AI systems with human values and societal norms, a challenge compounded by their increasing autonomy. The safety and robustness of AGI systems in diverse environments also remain a significant research gap. Addressing these challenges requires a multidisciplinary approach, incorporating ethical, social, and philosophical perspectives. Our survey has highlighted key areas for future inter-disciplinary research in AI, emphasizing the integration of ethical, sociological, and technical perspectives. This approach will foster collaborative research, bridging the gap between technological advancement and societal needs, ensuring that AI development is aligned with human values and global welfare. The roles of MoE, multimodal, and AGI in reshaping generative AI have been identified as significant, as their advancements can enhance model performance and versatility, and pave the way for future research in areas like ethical AI alignment and AGI. As we forge ahead, the balance between AI advancements and human creativity is not just a goal but a necessity, ensuring AI’s role as a complementary force that amplifies our capacity to innovate and solve complex challenges. Our responsibility is to guide these advancements towards enriching the human experience, aligning technological progress with ethical standards and societal well-being. ',\n",
" 'in_training': False}"
"dict_keys(['title', 'url', 'content'])"
]
},
"execution_count": 4,
"execution_count": 3,
"metadata": {},
"output_type": "execute_result"
}
],
"source": [
"MAX_LEN = 2000\n",
"import json\n",
"samples = json.load(open(\"../samples.json\"))\n",
"df_samples = pd.DataFrame(samples)\n",
"df_samples['len'] = df_samples['text'].str.len()\n",
"df_samples\n",
"\n",
"\n",
"sample = samples[0]\n",
"sample"
"import frontmatter\n",
"from pathlib import Path\n",
"sample_files = sorted(Path(\"../samples/\").glob('*.md'))\n",
"print(sample_files)\n",
"samples = [frontmatter.load(f).to_dict() for f in sample_files]\n",
"samples[0].keys()"
]
},
{
@@ -93,7 +94,7 @@
},
{
"cell_type": "code",
"execution_count": 5,
"execution_count": 4,
"metadata": {},
"outputs": [
{
@@ -102,7 +103,7 @@
"True"
]
},
"execution_count": 5,
"execution_count": 4,
"metadata": {},
"output_type": "execute_result"
}
@@ -121,12 +122,20 @@
{
"data": {
"text/plain": [
"'Research on representation engineering (RepE) for AI systems revealed new insights for monitoring and control. New methods were proposed, showing potential for safety-related issues. Future work could explore other aspects of AI representations.'"
"'Louise Pentland, former PayPal Chief Business Affairs and Legal Officer, has been appointed as Chief Counsel for Disney Parks, Experiences and Products, effective from September 15, 2023.'"
]
},
"execution_count": 6,
"metadata": {},
"output_type": "execute_result"
},
{
"ename": "",
"evalue": "",
"output_type": "error",
"traceback": [
"\u001b[1;31mThe Kernel crashed while executing code in the the current cell or a previous cell. Please review the code in the cell(s) to identify a possible cause of the failure. Click <a href='https://aka.ms/vscodeJupyterKernelCrash'>here</a> for more info. View Jupyter <a href='command:jupyter.viewOutput'>log</a> for further details."
]
}
],
"source": [
@@ -149,13 +158,13 @@
" r = chat_completion.choices[0].message.content\n",
" return r\n",
"\n",
"r = summize(samples[1][\"text\"])\n",
"r = summize(samples[1][\"content\"])\n",
"r"
]
},
{
"cell_type": "code",
"execution_count": 7,
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
@@ -261,7 +270,7 @@
},
{
"cell_type": "code",
"execution_count": 8,
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
@@ -284,7 +293,7 @@
},
{
"cell_type": "code",
"execution_count": 9,
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
@@ -304,7 +313,7 @@
},
{
"cell_type": "code",
"execution_count": 10,
"execution_count": null,
"metadata": {},
"outputs": [],
"source": [
@@ -318,7 +327,7 @@
},
{
"cell_type": "code",
"execution_count": 11,
"execution_count": null,
"metadata": {},
"outputs": [
{
@@ -519,12 +528,12 @@
"for model_name in models:\n",
" model, tokenizer = load_model(model_name)\n",
" for sample in samples:\n",
" if sample['name'] not in summaries:\n",
" summaries[sample['name']] = summize(sample['text'])[:600]\n",
" summary = summaries[sample['name']]\n",
" if sample['title'] not in summaries:\n",
" summaries[sample['title']] = summize(sample['content'])[:600]\n",
" summary = summaries[sample['title']]\n",
"\n",
" # before \n",
" s1 = sample['text']\n",
" s1 = sample['content']\n",
" results = perplexity_compute(data=s1, model=model, tokenizer=tokenizer, device='cuda')\n",
" before = results['mean_perplexity']\n",
"\n",
@@ -533,27 +542,20 @@
" High level summary: {summary}\n",
"\n",
"Text:\n",
"{sample['text']}\n",
"{sample['content']}\n",
" \"\"\"\n",
" results = perplexity_compute(data=s2, model=model, tokenizer=tokenizer, device='cuda')\n",
" after = np.array(results['perplexities'])[-len(s1):].mean()\n",
"\n",
" print(model_name, sample['name'], before, after)\n",
" data.append(dict(before=before, after=after, model=model_name, sample=sample['name'],\n",
" in_training=sample['in_training'], len=len(sample['text'])))\n"
" print(model_name, sample['title'], before, after)\n",
" data.append(dict(before=before, after=after, model=model_name, sample=sample['title'],\n",
" in_training=sample['in_training'], len=len(sample['content'])))\n"
]
},
{
"cell_type": "code",
"execution_count": null,
"metadata": {},
"outputs": [],
"source": []
},
{
"cell_type": "code",
"execution_count": 20,
"metadata": {},
"outputs": [
{
"data": {
@@ -719,7 +721,7 @@
},
{
"cell_type": "code",
"execution_count": 24,
"execution_count": null,
"metadata": {},
"outputs": [
{
poetry.lock
Generated
+19 -1
View File
@@ -2594,6 +2594,24 @@ files = [
[package.extras]
cli = ["click (>=5.0)"]
[[package]]
name = "python-frontmatter"
version = "1.0.1"
description = "Parse and manage posts with YAML (or other) frontmatter"
optional = false
python-versions = "*"
files = [
{file = "python-frontmatter-1.0.1.tar.gz", hash = "sha256:a6a082844fc601f34e4dd576bed8fcb5ef19112166e087629e4d6ba9bf4f7c35"},
{file = "python_frontmatter-1.0.1-py3-none-any.whl", hash = "sha256:0599198cc01b445e5d0be74ff35be0a6c7442dddbdb0803e018be4e055397f6a"},
]
[package.dependencies]
PyYAML = "*"
[package.extras]
docs = ["sphinx"]
test = ["pyaml", "pytest", "toml"]
[[package]]
name = "pytorch-lightning"
version = "2.1.3"
@@ -3960,4 +3978,4 @@ multidict = ">=4.0"
[metadata]
lock-version = "2.0"
python-versions = ">=3.10,<3.13"
content-hash = "fc4925d74f716ebc1454d911c30890cf2591d6881a28e1241bd8058032e38e87"
content-hash = "2856534c176a36679a1d86e5fd77f1008b5084ee100b22102d64a7c71eff7448"
pyproject.toml
+1
View File
@@ -27,6 +27,7 @@ einops = "^0.7.0"
tabulate = "^0.9.0"
lightning = "^2.1.3"
matplotlib = "^3.8.0"
python-frontmatter = "^1.0.1"
[[tool.poetry.source]]
name = "pytorch"
research_log.md
+6
View File
@@ -0,0 +1,6 @@
# 2024-01-03 18:57:01
Initial version
- added large docs. But not the results don't make sense. I think I have a problem where the first and latter half of the docs are diff
- [ ] I need to window, then spit into train and test
samples.json
-1
View File
File diff suppressed because one or more lines are too long
samples/bletchley_decleration.md
+28
View File
@@ -0,0 +1,28 @@
---
title: blechley declaration
url: https://www.industry.gov.au/publications/bletchley-declaration-countries-attending-ai-safety-summit-1-2-november-2023
---
Text of the Bletchley Declaration
Artificial Intelligence (AI) presents enormous global opportunities: it has the potential to transform and enhance human wellbeing, peace and prosperity. To realise this, we affirm that, for the good of all, AI should be designed, developed, deployed, and used, in a manner that is safe, in such a way as to be human-centric, trustworthy and responsible. We welcome the international community’s efforts so far to cooperate on AI to promote inclusive economic growth, sustainable development and innovation, to protect human rights and fundamental freedoms, and to foster public trust and confidence in AI systems to fully realise their potential.
AI systems are already deployed across many domains of daily life including housing, employment, transport, education, health, accessibility, and justice, and their use is likely to increase. We recognise that this is therefore a unique moment to act and affirm the need for the safe development of AI and for the transformative opportunities of AI to be used for good and for all, in an inclusive manner in our countries and globally. This includes for public services such as health and education, food security, in science, clean energy, biodiversity, and climate, to realise the enjoyment of human rights, and to strengthen efforts towards the achievement of the United Nations Sustainable Development Goals.
Alongside these opportunities, AI also poses significant risks, including in those domains of daily life. To that end, we welcome relevant international efforts to examine and address the potential impact of AI systems in existing fora and other relevant initiatives, and the recognition that the protection of human rights, transparency and explainability, fairness, accountability, regulation, safety, appropriate human oversight, ethics, bias mitigation, privacy and data protection needs to be addressed. We also note the potential for unforeseen risks stemming from the capability to manipulate content or generate deceptive content. All of these issues are critically important and we affirm the necessity and urgency of addressing them.
Particular safety risks arise at the ‘frontier’ of AI, understood as being those highly capable general-purpose AI models, including foundation models, that could perform a wide variety of tasks - as well as relevant specific narrow AI that could exhibit capabilities that cause harm - which match or exceed the capabilities present in today’s most advanced models. Substantial risks may arise from potential intentional misuse or unintended issues of control relating to alignment with human intent. These issues are in part because those capabilities are not fully understood and are therefore hard to predict. We are especially concerned by such risks in domains such as cybersecurity and biotechnology, as well as where frontier AI systems may amplify risks such as disinformation. There is potential for serious, even catastrophic, harm, either deliberate or unintentional, stemming from the most significant capabilities of these AI models. Given the rapid and uncertain rate of change of AI, and in the context of the acceleration of investment in technology, we affirm that deepening our understanding of these potential risks and of actions to address them is especially urgent.
Many risks arising from AI are inherently international in nature, and so are best addressed through international cooperation. We resolve to work together in an inclusive manner to ensure human-centric, trustworthy and responsible AI that is safe, and supports the good of all through existing international fora and other relevant initiatives, to promote cooperation to address the broad range of risks posed by AI. In doing so, we recognise that countries should consider the importance of a pro-innovation and proportionate governance and regulatory approach that maximises the benefits and takes into account the risks associated with AI. This could include making, where appropriate, classifications and categorisations of risk based on national circumstances and applicable legal frameworks. We also note the relevance of cooperation, where appropriate, on approaches such as common principles and codes of conduct. With regard to the specific risks most likely found in relation to frontier AI, we resolve to intensify and sustain our cooperation, and broaden it with further countries, to identify, understand and as appropriate act, through existing international fora and other relevant initiatives, including future international AI Safety Summits.
All actors have a role to play in ensuring the safety of AI: nations, international fora and other initiatives, companies, civil society and academia will need to work together. Noting the importance of inclusive AI and bridging the digital divide, we reaffirm that international collaboration should endeavour to engage and involve a broad range of partners as appropriate, and welcome development-orientated approaches and policies that could help developing countries strengthen AI capacity building and leverage the enabling role of AI to support sustainable growth and address the development gap.
We affirm that, whilst safety must be considered across the AI lifecycle, actors developing frontier AI capabilities, in particular those AI systems which are unusually powerful and potentially harmful, have a particularly strong responsibility for ensuring the safety of these AI systems, including through systems for safety testing, through evaluations, and by other appropriate measures. We encourage all relevant actors to provide context-appropriate transparency and accountability on their plans to measure, monitor and mitigate potentially harmful capabilities and the associated effects that may emerge, in particular to prevent misuse and issues of control, and the amplification of other risks.
In the context of our cooperation, and to inform action at the national and international levels, our agenda for addressing frontier AI risk will focus on:
identifying AI safety risks of shared concern, building a shared scientific and evidence-based understanding of these risks, and sustaining that understanding as capabilities continue to increase, in the context of a wider global approach to understanding the impact of AI in our societies.
building respective risk-based policies across our countries to ensure safety in light of such risks, collaborating as appropriate while recognising our approaches may differ based on national circumstances and applicable legal frameworks. This includes, alongside increased transparency by private actors developing frontier AI capabilities, appropriate evaluation metrics, tools for safety testing, and developing relevant public sector capability and scientific research.
In furtherance of this agenda, we resolve to support an internationally inclusive network of scientific research on frontier AI safety that encompasses and complements existing and new multilateral, plurilateral and bilateral collaboration, including through existing international fora and other relevant initiatives, to facilitate the provision of the best science available for policy making and the public good.
In recognition of the transformative positive potential of AI, and as part of ensuring wider international cooperation on AI, we resolve to sustain an inclusive global dialogue that engages existing international fora and other relevant initiatives and contributes in an open manner to broader international discussions, and to continue research on frontier AI safety to ensure that the benefits of the technology can be harnessed responsibly for good and for all. We look forward to meeting again in 2024.
samples/cicero_fin1.md
+33
View File
@@ -0,0 +1,33 @@
---
title: cicero from ibois, Philippe (2012-06-03).
url: http://www.thelatinlibrary.com/cicero/fin1.shtml
---
M. TVLLI CICERONIS DE FINIBVS BONORVM ET MALORVM LIBER PRIMVS
[1] Non eram nescius, Brute, cum, quae summis ingeniis exquisitaque doctrina philosophi Graeco sermone tractavissent, ea Latinis litteris mandaremus, fore ut hic noster labor in varias reprehensiones incurreret. nam quibusdam, et iis quidem non admodum indoctis, totum hoc displicet philosophari. quidam autem non tam id reprehendunt, si remissius agatur, sed tantum studium tamque multam operam ponendam in eo non arbitrantur. erunt etiam, et ii quidem eruditi Graecis litteris, contemnentes Latinas, qui se dicant in Graecis legendis operam malle consumere. postremo aliquos futuros suspicor, qui me ad alias litteras vocent, genus hoc scribendi, etsi sit elegans, personae tamen et dignitatis esse negent. [2] Contra quos omnis dicendum breviter existimo. Quamquam philosophiae quidem vituperatoribus satis responsum est eo libro, quo a nobis philosophia defensa et collaudata est, cum esset accusata et vituperata ab Hortensio. qui liber cum et tibi probatus videretur et iis, quos ego posse iudicare arbitrarer, plura suscepi veritus ne movere hominum studia viderer, retinere non posse. Qui autem, si maxime hoc placeat, moderatius tamen id volunt fieri, difficilem quandam temperantiam postulant in eo, quod semel admissum coerceri reprimique non potest, ut propemodum iustioribus utamur illis, qui omnino avocent a philosophia, quam his, qui rebus infinitis modum constituant in reque eo meliore, quo maior sit, mediocritatem desiderent. [3] Sive enim ad sapientiam perveniri potest, non paranda nobis solum ea, sed fruenda etiam [sapientia] est; sive hoc difficile est, tamen nec modus est ullus investigandi veri, nisi inveneris, et quaerendi defatigatio turpis est, cum id, quod quaeritur, sit pulcherrimum. etenim si delectamur, cum scribimus, quis est tam invidus, qui ab eo nos abducat? sin laboramus, quis est, qui alienae modum statuat industriae? nam ut Terentianus Chremes non inhumanus, qui novum vicinum non vult 'fodere aut arare aut aliquid ferre denique' -- non enim illum ab industria, sed ab inliberali labore deterret --, sic isti curiosi, quos offendit noster minime nobis iniucundus labor.
[4] Iis igitur est difficilius satis facere, qui se Latina scripta dicunt contemnere. in quibus hoc primum est in quo admirer, cur in gravissimis rebus non delectet eos sermo patrius, cum idem fabellas Latinas ad verbum e Graecis expressas non inviti legant. quis enim tam inimicus paene nomini Romano est, qui Ennii Medeam aut Antiopam Pacuvii spernat aut reiciat, quod se isdem Euripidis fabulis delectari dicat, Latinas litteras oderit?
Synephebos ego, inquit, potius Caecilii aut Andriam Terentii quam utramque Menandri legam? [5] A quibus tantum dissentio, ut, cum Sophocles vel optime scripserit Electram, tamen male conversam Atilii mihi legendam putem, de quo Lucilius: 'ferreum scriptorem', verum, opinor, scriptorem tamen, ut legendus sit. rudem enim esse omnino in nostris poetis aut inertissimae segnitiae est aut fastidii delicatissimi. mihi quidem nulli satis eruditi videntur, quibus nostra ignota sunt. an 'Utinam ne in nemore . . .' nihilo minus legimus quam hoc idem Graecum, quae autem de bene beateque vivendo a Platone disputata sunt, haec explicari non placebit Latine? [6] Quid? si nos non interpretum fungimur munere, sed tuemur ea, quae dicta sunt ab iis quos probamus, eisque nostrum iudicium et nostrum scribendi ordinem adiungimus, quid habent, cur Graeca anteponant iis, quae et splendide dicta sint neque sint conversa de Graecis? nam si dicent ab illis has res esse tractatas, ne ipsos quidem Graecos est cur tam multos legant, quam legendi sunt. quid enim est a Chrysippo praetermissum in Stoicis? legimus tamen Diogenem, Antipatrum, Mnesarchum, Panaetium, multos alios in primisque familiarem nostrum Posidonium. quid? Theophrastus mediocriterne delectat, cum tractat locos ab Aristotele ante tractatos? quid? Epicurei num desistunt de isdem, de quibus et ab Epicuro scriptum est et ab antiquis, ad arbitrium suum scribere? quodsi Graeci leguntur a Graecis isdem de rebus alia ratione compositis, quid est, cur nostri a nostris non legantur?
[7] Quamquam, si plane sic verterem Platonem aut Aristotelem, ut verterunt nostri poetae fabulas, male, credo, mererer de meis civibus, si ad eorum cognitionem divina illa ingenia transferrem. sed id neque feci adhuc nec mihi tamen, ne faciam, interdictum puto. locos quidem quosdam, si videbitur, transferam, et maxime ab iis, quos modo nominavi, cum inciderit, ut id apte fieri possit, ut ab Homero Ennius, Afranius a Menandro solet. Nec vero, ut noster Lucilius, recusabo, quo minus omnes mea legant. utinam esset ille Persius, Scipio vero et Rutilius multo etiam magis, quorum ille iudicium reformidans Tarentinis ait se et Consentinis et Siculis scribere. facete is quidem, sicut alia; sed neque tam docti tum erant, ad quorum iudicium elaboraret, et sunt illius scripta leviora, ut urbanitas summa appareat, doctrina mediocris. [8] Ego autem quem timeam lectorem, cum ad te ne Graecis quidem cedentem in philosophia audeam scribere? quamquam a te ipso id quidem facio provocatus gratissimo mihi libro, quem ad me de virtute misisti. Sed ex eo credo quibusdam usu venire; ut abhorreant a Latinis, quod inciderint in inculta quaedam et horrida, de malis Graecis Latine scripta deterius. quibus ego assentior, dum modo de isdem rebus ne Graecos quidem legendos putent. res vero bonas verbis electis graviter ornateque dictas quis non legat? nisi qui se plane Graecum dici velit, ut a Scaevola est praetore salutatus Athenis Albucius. [9] Quem quidem locum comit multa venustate et omni sale idem Lucilius, apud quem praeclare Scaevola:
Graecum te, Albuci, quam Romanum atque Sabinum,
municipem Ponti, Tritani, centurionum,
praeclarorum hominum ac primorum signiferumque,
maluisti dici. Graece ergo praetor Athenis,
id quod maluisti, te, cum ad me accedis, saluto:
'chaere,' inquam, 'Tite!' lictores, turma omnis chorusque:
'chaere, Tite!' hinc hostis mi Albucius, hinc inimicus.
[10] Sed iure Mucius. ego autem mirari [satis] non queo unde hoc sit tam insolens domesticarum rerum fastidium. non est omnino hic docendi locus; sed ita sentio et saepe disserui, Latinam linguam non modo non inopem, ut vulgo putarent, sed locupletiorem etiam esse quam Graecam. quando enim nobis, vel dicam aut oratoribus bonis aut poetis, postea quidem quam fuit quem imitarentur, ullus orationis vel copiosae vel elegantis ornatus defuit? Ego vero, quoniam forensibus operis, laboribus, periculis non deseruisse mihi videor praesidium, in quo a populo Romano locatus sum, debeo profecto, quantumcumque possum, in eo quoque elaborare, ut sint opera, studio, labore meo doctiores cives mei, nec cum istis tantopere pugnare, qui Graeca legere malint, modo legant illa ipsa, ne simulent, et iis servire, qui vel utrisque litteris uti velint vel, si suas habent, illas non magnopere desiderent. [11] Qui autem alia malunt scribi a nobis, aequi esse debent, quod et scripta multa sunt, sic ut plura nemini e nostris, et scribentur fortasse plura, si vita suppetet; et tamen, qui diligenter haec, quae de philosophia litteris mandamus, legere assueverit, iudicabit nulla ad legendum his esse potiora. quid est enim in vita tantopere quaerendum quam cum omnia in philosophia, tum id, quod his libris quaeritur, qui sit finis, quid extremum, quid ultimum, quo sint omnia bene vivendi recteque faciendi consilia referenda, quid sequatur natura ut summum ex rebus expetendis, quid fugiat ut extremum malorum? qua de re cum sit inter doctissimos summa dissensio, quis alienum putet eius esse dignitatis, quam mihi quisque tribuat, quid in omni munere vitae optimum et verissimum sit, exquirere? [12] An, partus ancillae sitne in fructu habendus, disseretur inter principes civitatis, P. Scaevolam M'.que Manilium, ab iisque M. Brutus dissentiet -- quod et acutum genus est et ad usus civium non inutile, nosque ea scripta reliquaque eiusdem generis et legimus libenter et legemus --, haec, quae vitam omnem continent, neglegentur? nam, ut sint illa vendibiliora, haec uberiora certe sunt. quamquam id quidem licebit iis existimare, qui legerint. nos autem hanc omnem quaestionem de finibus bonorum et malorum fere a nobis explicatam esse his litteris arbitramur, in quibus, quantum potuimus, non modo quid nobis probaretur, sed etiam quid a singulis philosophiae disciplinis diceretur, persecuti sumus.
[13] Ut autem a facillimis ordiamur, prima veniat in medium Epicuri ratio, quae plerisque notissima est. quam a nobis sic intelleges eitam, ut ab ipsis, qui eam disciplinam probant, non soleat accuratius explicari; verum enim invenire volumus, non tamquam adversarium aliquem convincere. accurate autem quondam a L. Torquato, homine omni doctrina erudito, defensa est Epicuri sententia de voluptate, a meque ei responsum, cum C. Triarius, in primis gravis et doctus adolescens, ei disputationi interesset. [14] Nam cum ad me in Cumanum salutandi causa uterque venisset, pauca primo inter nos de litteris, quarum summum erat in utroque studium, deinde Torquatus: Quoniam nacti te, inquit, sumus aliquando otiosum, certe audiam, quid sit, quod Epicurum nostrum non tu quidem oderis, ut fere faciunt, qui ab eo dissentiunt, sed certe non probes, eum quem ego arbitror unum vidisse verum maximisque erroribus animos hominum liberavisse et omnia tradidisse, quae pertinerent ad bene beateque vivendum. sed existimo te, sicut nostrum Triarium, minus ab eo delectari, quod ista Platonis, Aristoteli, Theophrasti orationis ornamenta neglexerit. nam illud quidem adduci vix possum, ut ea, quae senserit ille, tibi non vera videantur.
[15] Vide, quantum, inquam, fallare, Torquate. oratio me istius philosophi non offendit; nam et complectitur verbis, quod vult, et dicit plane, quod intellegam; et tamen ego a philosopho, si afferat eloquentiam, non asperner, si non habeat, non admodum flagitem. re mihi non aeque satisfacit, et quidem locis pluribus. sed quot homines, tot sententiae; falli igitur possumus.
Quam ob rem tandem, inquit, non satisfacit? te enim iudicem aequum puto, modo quae dicat ille bene noris.
[16] Nisi mihi Phaedrum, inquam, tu mentitum aut Zenonem putas, quorum utrumque audivi, cum mihi nihil sane praeter sedulitatem probarent, omnes mihi Epicuri sententiae satis notae sunt. atque eos, quos nominavi, cum Attico nostro frequenter audivi, cum miraretur ille quidem utrumque, Phaedrum autem etiam amaret, cotidieque inter nos ea, quae audiebamus, conferebamus, neque erat umquam controversia, quid ego intellegerem, sed quid probarem.
[17] Quid igitur est? inquit; audire enim cupio, quid non probes. Principio, inquam, in physicis, quibus maxime gloriatur, primum totus est alienus. Democritea dicit perpauca mutans, sed ita, ut ea, quae corrigere vult, mihi quidem depravare videatur. ille atomos quas appellat, id est corpora individua propter soliditatem, censet in infinito inani, in quo nihil nec summum nec infimum nec medium nec ultimum nec extremum sit, ita ferri, ut concursionibus inter se cohaerescant, ex quo efficiantur ea, quae sint quaeque cernantur, omnia, eumque motum atomorum nullo a principio, sed ex aeterno tempore intellegi convenire. [18] Epicurus autem, in quibus sequitur Democritum, non fere labitur. quamquam utriusque cum multa non probo, tum illud in primis, quod, cum in rerum natura duo quaerenda sint, unum, quae materia sit, ex qua quaeque res efficiatur, alterum, quae vis sit, quae quidque efficiat, de materia disseruerunt, vim et causam efficiendi reliquerunt. sed hoc commune vitium, illae Epicuri propriae ruinae: censet enim eadem illa individua et solida corpora ferri deorsum suo pondere ad lineam, hunc naturalem esse omnium corporum motum. [19] Deinde ibidem homo acutus, cum illud ocurreret, si omnia deorsus e regione ferrentur et, ut dixi, ad lineam, numquam fore ut atomus altera alteram posset attingere itaque ** attulit rem commenticiam: declinare dixit atomum perpaulum, quo nihil posset fieri minus; ita effici complexiones et copulationes et adhaesiones atomorum inter se, ex quo efficeretur mundus omnesque partes mundi, quaeque in eo essent. Quae cum tota res (est) ficta pueriliter, tum ne efficit [quidem], quod vult. nam et ipsa declinatio ad libidinem fingitur -- ait enim declinare atomum sine causa; quo nihil turpius physico, quam fieri quicquam sine causa dicere, -- et illum motum naturalem omnium ponderum, ut ipse constituit, e regione inferiorem locum petentium sine causa eripuit atomis nec tamen id, cuius causa haec finxerat, assecutus est. [20] Nam si omnes atomi declinabunt, nullae umquam cohaerescent, sive aliae declinabunt, aliae suo nutu recte ferentur, primum erit hoc quasi, provincias atomis dare, quae recte, quae oblique ferantur, deinde eadem illa atomorum, in quo etiam Democritus haeret, turbulenta concursio hunc mundi ornatum efficere non poterit. ne illud quidem physici, credere aliquid esse minimum, quod profecto numquam putavisset, si a Polyaeno, familiari suo, geometrica discere maluisset quam illum etiam ipsum dedocere. Sol Democrito magnus videtur, quippe homini erudito in geometriaque perfecto, huic pedalis fortasse; tantum enim esse censet, quantus videtur, vel paulo aut maiorem aut minorem. [21] Ita, quae mutat, ea corrumpit, quae sequitur sunt tota Democriti, atomi, inane, imagines, quae eidola nominant, quorum incursione non solum videamus, sed etiam cogitemus; infinitio ipsa, quam apeirian vocant, tota ab illo est, tum innumerabiles mundi, qui et oriantur et intereant cotidie. Quae etsi mihi nullo modo probantur, tamen Democritum laudatum a ceteris ab hoc, qui eum unum secutus esset, nollem vituperatum.
samples/disney_appointment.md
+31
View File
@@ -0,0 +1,31 @@
---
title: disney appointment
url: https://thewaltdisneycompany.com/louise-pentland-named-chief-counsel-for-disney-parks-experiences-and-products/
---
BURBANK, Calif., September 6, 2023 — Louise Pentland has been named Chief Counsel for the Disney Parks, Experiences and Products segment of The Walt Disney Company (NYSE: DIS), effective September 15. Pentland, who previously served as Chief Business Affairs and Legal Officer at PayPal, will report to Horacio Gutierrez, Senior Executive Vice President, General Counsel and Chief Compliance Officer of The Walt Disney Company, and act as a strategic advisor to Josh D’Amaro, Chairman, Disney Parks, Experiences and Products and his executive leadership team.
In this role, Pentland will oversee all legal and regulatory matters for the segment’s businesses, including six theme park-resort destinations in the United States, Europe and Asia; a top-rated cruise line; a popular vacation ownership program; an award-winning guided family adventure business; and Disney’s global consumer products operations, which include the world’s leading licensing business across toys, apparel, home goods, digital games and apps; the world’s largest children’s print publisher; and the shopDisney e-commerce platform.
“Louise Pentland is a highly accomplished legal executive with more than two decades of experience serving at the highest levels of innovative global companies,” said Gutierrez. “Her broad expertise working with complex business, regulatory and compliance matters around the world and her track record as an exemplary leader make her a superb addition to the leadership team at Disney Parks, Experiences and Products.”
“This is an exciting time to join The Walt Disney Company and I’m honored to be part of a business that brings so much joy to people across the globe – myself included,” said Pentland. “I’ve long admired Disney Parks’ reputation as a global leader in tourism and consumer products, along with its exceptional history of innovation and growth, and I’m looking forward to taking on this role and helping advance the organization’s ambitious goals for the future.”
At PayPal, Pentland served most recently as EVP, Chief Business Affairs and Legal Officer, and was responsible for the company’s global People, Legal, Communications, Intellectual Property, Government Relations and Social Innovation functions, with oversight over all financial, regulatory and legal requirements worldwide. She led stakeholder engagement to promote PayPal’s reputation globally through corporate affairs initiatives and oversaw the company’s People function, including diversity and inclusion initiatives, workforce planning, employee relations, compensation, benefits, leadership excellence, culture, pro-bono initiatives and ESG.
Prior to PayPal, Pentland was Executive Vice President and Chief Legal Officer at Nokia Corporation, where she directed the company’s legal and government affairs functions, M&A and IP management, for which her team was named Europe’s most innovative in-house legal team by the Financial Times in 2011.
Pentland has served as a member of Hitachi Ltd.’s Board of Directors since 2015, Pacific Life’s Board of Directors since 2020 and Experian PLC’s Board of Directors since 2022. Additionally, she was named to the Board of Directors of the non-profit Black Executive CMO Alliance (BECA) in 2021.
Pentland holds an undergraduate law degree from Northumbria University, Newcastle upon Tyne, and a Postgraduate Diploma in Legal Practice from Northumbria University. Pentland is licensed to practice law in the U.S., England, Wales and Europe.
Contacts:
David Jefferson
Corporate Communications
david.j.jefferson@disney.com
(818) 560-4832
Mike Long
Corporate Communications
mike.p.long@disney.com
(818) 560-4588
samples/fake_paper.md
+42
View File
@@ -0,0 +1,42 @@
---
title: fake ai hoax paper
url:
---
Title: "Deconstructing Binaries: Interrogating the Hegemonic Paradigms in Machine Learning"
Abstract:
Machine Learning (ML) as a field has largely embraced binary constructs as foundational to its functioning, with 0/1 and true/false distinctions underpinning many of its core algorithms. This paper argues that these binary frameworks are not merely technical conveniences but are, in fact, reflective of deeper hegemonic paradigms that perpetuate exclusionary practices and systemic biases. By uncritically adopting these binaries, ML inadvertently reinforces a worldview that marginalizes complex, nuanced identities and experiences.
We propose a critical examination of these binaries, questioning the necessity and ubiquity of dualistic thinking within ML. We suggest that the field's reliance on binary classification not only limits its predictive accuracy in certain contexts but also fails to capture the rich, fluid nature of human experiences and societal structures. Instead, we advocate for a "fluid" approach to algorithms that allows for more nuanced and inclusive representations of reality, rejecting the oversimplified and often exclusionary nature of strict binary outcomes.
Furthermore, this paper argues for the integration of intersectional data that reflects the diverse and overlapping categories of identity, including race, gender, class, and more. Current ML models often overlook these complexities, leading to outcomes that fail to serve, and even harm, underrepresented populations. We critique the prevailing notion that ML is a neutral, objective tool, highlighting the lack of socio-political context in algorithmic decision-making processes.
In conclusion, we call for an epistemological shift in the field of Machine Learning. This shift involves moving away from a purely positivistic, binary approach towards one that is reflective, inclusive, and aware of the social dimensions of technology. By reimagining the foundational paradigms of ML, we can work towards a more equitable and nuanced understanding of the world, one that respects and represents the full spectrum of human experience.
Appendix:
A. Expanded Discussion on Binary Constructs in ML:
Detailed analysis of how binary constructs are embedded in ML algorithms.
Case studies highlighting the limitations and biases introduced by binary classification.
B. Theoretical Framework for a Fluid Approach:
Outline of theoretical underpinnings for moving beyond binary thinking.
Discussion on existing and potential methodologies that embrace complexity and fluidity.
C. Intersectionality in Data Science:
Exploration of intersectionality theory and its relevance to ML.
Examples of how intersectional analysis can be incorporated into data sets and model design.
D. Case Studies of Non-binary Models:
Detailed case studies where non-binary models have been successfully implemented.
Analysis of the outcomes, challenges, and opportunities presented by these models.
E. Ethical Considerations and Future Directions:
Discussion on the ethical implications of perpetuating binary constructs in ML.
Vision for a future where ML is more inclusive, reflective, and socially aware, including potential pathways and strategies to achieve this goal.
samples/fauci_emails.md
+299
View File
@@ -0,0 +1,299 @@
---
title: buzzfeed foi fauci emails 2023
url: https://s3.documentcloud.org/documents/20793561/leopold-nih-foia-anthony-fauci-emails.pdf
---
Dear Dr. Fauci: I am an ex-alumni from the Department of Transfusion Medicine at the NIH. I am sorry to
bother but I wanted to not ify the Coronavirus Task force about potential blood shortages if blood
centers get blood drive cancellations due to the fear in the public.
We are starting to see this locally and my concern is that other blood centers in the country are facing
the same issue.
I tried to contact the office of the vice president, and all I could do was to send an email, which might
get lost among thousands .
I am hoping you see this and perhaps br ing this to the table. Blood centers use updated federal
information when educating donors, staff and hospitals but the current fea r is tr iggering blood drive
cancellations. This might be experienced nationally and will impact the national blood invento ry.
Thank you
Maria De Los Angeles Muniz (Angie) , MD
Medical Director
o 201.389.0439 I c ._____ Cb_H_ (please call cell phone first)
mmun iz@v italant. org
Vitalant Mon tvale
Bank
102 Chestnut Ridge Road
Montva le, NJ 07645
V italant Cleveland Vitalant Cord Blood
333 East Bridge Street 1 Pearl Court
Elyria, OH 44035 Allendale, NJ 07401
Community Blood Serv ices, Lifeshare & Blood Systems are now Vi talant.
NIH-000964
From:
Sent:
Conrad, Patricia (NIH/NIAID) [El on behalf of Fauci, Anthony (NIH/NIAIO) [El
Thu, 5 Mar 2020 21:27:53 +0000
To: NIAIDODAM
Subject: FW: 03 04 2020 Or. Anthony Fauci RE Invite to join my head table at Global's
AcceptAbility Gala on Wed S/20
Attachment s: 2020 03 GLOBALAcceptAbility Gala Save the Date.pdf
Importance: High
Patricia L. Conrad
Public Health Ana lyst and
Special Assistant to the Direc tor
Nat ional Institute of A ller gy and Infectious Diseases
Th e National Institutes of Hea lth
31 Center Drive, MSC 2520 - Room 7A03
Bethesda, Maryland 20892
(b)(6)
30 1-496-4409 fax
Disclaimer :
The informat ion in this e-ma il and any of its attachments is confidentia l and may conta in sensi tive information. It should not be used
by anyone who is not the original intended recipient. If you have received this e-mail in error please inform the sender and delete it
from your mailbox or any other storage devices. Nationa l Institute of Allergy ancl Infectious Diseases (NIAID) shall not accept
liabil ity for any statement macle that are sencler"s own ancl not expressly macle on behalf of the NIAID by one of its representatives.
From: M ichelle Whitten - Global <mswhitten@gl oba ldownsynd rome.org>
Sent: Wednesday, March 4, 2020 10:35 PM
To: Fauci, Anthony (NIH/NIAID) [E]------ --=-(b=)(=6)
Cc: Ashley Sparhawk <asparhawk@globaldownsyndrome .org>; Marisa Cucuzzella
--------~ =<mcucuzzella@ajsfoundation .com>; Rotrosen, Daniel (NIH/NIA ID) [El (b)(6l;
Deckhut, Alison (NIH/NIAID) [El CbH6)>;Rothermel, Annette (NIH/N IAID) [El
(b) (6) >
Subject: 03 04 2020 Dr. Anthony Fauci RE Invite to jo in my head table at Global's AcceptAbi lity Gala on
Wed S/20
Importance: High
Dear Dr. Fauci,
First let me say we are very grateful to everyone at NIH for working to contain the coronavirus,
especially your team at the NIAID. Your interviews are articulate and informative - they give Americans
some comfort and some concrete ways to t hink about protecting themselves and what to expect. It
must be a stressful ti me and we want you to know that we have confidence in our NIH and CDCto
contain it and to help stave it off globally as well.
NIH-000965
On a happier note, I wanted to reach out and THANK YOU for all you do for our Down Syndrome
research through the transformative trans-NIH INCLUDEprogram. It is making a HUGE difference!!
To this end, I am writing today in hopes you will be able to join us on Wednesday, May 20th for our
AcceptAbili ty Gala at my head table. Attached is our save the date for the event . We are excited to have
Caroline Cardenas as our 2020 Ambassador, Rep Pete Stauber as our keynote (b)(6)
•--• , and Reps Lucille Roybal-Allard and Jaime Herrera Beutler are Global's Quincy Jones
Exceptional Advocacy Awardees. We believe the gala will be quite lovely and meaningfu l to so many .
We do understand that attendance cannot be considered if the coronavirus situation worsens. However,
if things are look ing much better (touch wood) we wou ld be deeply honored if you and our friends from
NIAID could attend. Looking forward to hearing from you soon.
All the Best, Michelle and family
NIH-000966
From:
Sent :
To :
Cc:
Fauci, Anthony (NIH/NIAID) [E)
Thu, 5 Mar 2020 11:00:39 +0000
(b)(6)
Conrad, Patricia (NIH/NIAID) [E);Greg Folkers
-------~=________ Cb)_C_6);Marston, Hilary (NIH/NIAID) [E)
Subject: FW: New concerns for our randomized trial
What do you think? I see the FDACommissioner every day in person.
From: Kalil, Andre C (b)(6)>
Sent: Wednesday, March 4, 2020 11:27 PM
To: Seigel, John (NIH) (El (b)(6)•>; Davey, Richard (NIH/N IAID) [E]
(b)(6)>; Lane, Cliff (NIH/NIAID) [E] (b)(6); Marston, Hilary {NIH/NIAID)
~======~~[E] (b)(6)>; Fauci, Anthony (NIH/NIAID) [E] (b)(6)>
Subject: New concerns for our randomized trial
I want to share with you my concerns before the situation gets worse.
I have received many calls from physicians from all over the country about how to get
"compassionate use" remdesivir. (b)(4).(b)(5)
(b) (4). (b) (5)
I would appreciate hearing your thoughts on this critical issue.
NIH-000967
Best,
Andre
Andre Kalil, MD, MPH , FACP , FIDSA , FCCM
Professor
Departmentof Internal Medicine
Division of InfectiousDiseases
Director, Transplant ID Program
Associate Editor, CMI, Official Journal of ESCMID
Editorial Board, CCM, Official Journal of SCCM
~ UNMc ·
~ rcitlflL •
University of Nebraska Med ical Center
985400 Nebraska Medical Center, Omaha, NE 68198-5400
=== (b=)( ~6)~1=fax 402.559-5581
(b)(6)
UNMC I Facebook I Tw itter I YouTube I Flickr
"Lucky? Obv iously you haven 't heard anyth ing I've said . It was a matter of app lying Bayes' Theo rem to
est imate the condit ional probab ilities . Giving due we ight to the prio r probabi lities ..."
Robert Ludlum - The Amble r Warn ing
The informat ion in th is e-mail may be privileged and confidentia l, intended only for the use of the
addressee(s) above. Any unauthorized use or disclosure of this information is prohibited. If you have
received this e-ma il by mistake, please delete it and immediately contact the sender.
NIH-000968
From:
Sent:
To:
Subject :
Dave:
Fauci,Anthony (NIH/NIAID) [E)
Thu, 5 Mar 2020 03:29:00 +0000
Dave Doern
RE:Go Tony
Many thanks for your kind note. I hope that all is well with you.
Best regards ,
Tony
-----Original Message- ----
From: Dave Doern.=-------- (b=)~(=6)
Sent: Wednesday, March 4, 2020 12:l..9-'-P.;;...M.;;......___ ~~
To: Fauci, Anthony (NTH/NIAID) [E] (b)(6) ; Fauci, Anthony (NIH/NIA ID) [E]
(b)(6)>
Subject: Go Tony
Tony,
You always do the right thing ......comforting for all your
coronavirus attack ....
We'll do the praying and you go Keep doing the liard work .... .
Thanks,
Dave
DavidDoem
Sent from my iPhone
(b)(6)
NIH-000969
From:
Sent :
To:
Cc:
Subject:
Fauci, Anthony (NIH/NIAID) [E)
Thu, 5 Mar 2020 03:19:47 +0000
(b)(6)
Greg Folkers Cb)<)
---------RE: Conference Planning with respect to coronavirus.
There is no way of knowing for sure. I would wait unti l May and see what the dynamics of the
outbreak are globally and make your decision then whether or not to cancel.
From : (b)(6)
Sent : Wednesday, March 4, 2020 5:46 PM
------- ~=To : Fauci, Anthony (NIH/NIAID) [E (b)(6)
Subject : Conference Planning wit h respect to coronavirus.
Dear Dr. Fauci
Dr. Miriam Ke lty forme rly of NIH referred me to you . I am on the board of d irectors and planning
committee of the Applied Superconductivity Conference that is planned to be held in Tampa, Florida the
last week of July. Over 50% of our expected 1500 attendees are from outside the continental US. Our
planning committee appears to have three options :
1. Hold the conference on the original dates
2. Postpone the conference a few months
3. Cancel the confe rence
It would be most helpful to decision process if you can give us a prediction (anonymously of co urse) of
how the effects of the virus will pan out.
I look forward to your reply .
Bruce Strauss
ScD, MBA, PE, F-IEEE
NIH-000970
From:
Sent:
To :
Subject:
Fauci,Anthony (NIH/NIAIO) [E)
Thu, 5 Mar 2020 02:59:22 +0000
Aliantha Angel
RE:A humble request for your wisdom
From: Aliantha Angel (b)(6)
Sent: Wednesday, March 4, 2020 9:55 PM
To: Fauci,Anthony (NIH/NIAID) [E) ----- --:a(b=H=6)>
Subject : Re: A humble request for your wisdom
Oh my God ...
I honestly never expected you to reply and I thank you from the bottom of my heart for being so
generous!
Is there anything I can do for you besides being grateful?
You and yours are in my prayers!
much love,
Aliantha
On Wed, Mar 4, 2020 at 9:45 PM Fauci, Anth ony (NIH/N IAID) [E) (b) ~ > wrote :
--------
Dear Ms. Angel:
(b)(6)
(b)(6)
The severe complication of coronavirus are heavily skewed towards the elderly and those
with underlying conditions (Heart disease, Chronic lung disease, kidney disease, diabetes,
etc.) Most of the pneumon ias are pu re viral pneumonia and so this vaccination w ill not help
that. However, on the chance that you have a pure viral pneumonia that gets secondari ly
complicated by a bacterial pneumonia (pneumococcal) the vaccine would be beneficial. If
you are 65 years of age or older, you shou ld get the pneumonvax23 vaccine anyway
regardless of the risk of coronavirus infection.
Thanks,
Tony
From: Aliantha Angel (b)(6)
Sent: Wednesday, March 4, 2020 8:44 PM
NIH-00097 1
(b)(6)>To: Fauci, Anthony (NIH/NIAID) [E]--------Subject: A humble request for your wisdom
Good evening!
I know you must be completely busy and inundated with people want ing your time. I apologize that I
have nothing to offer in return and completely understand if you don't have time to answer. I called
the CDCbut they were totally unhelpful. I have a question that makes sense to me and I was hoping
you could answer and the answer might help a lot of people.
I understand that over time I, and everyone else, will very likely get COVI0-19 and that most people
won't even realize it because it will be minor. I get that, so th is is not a panicked q uestion.
I also understand that while most cases wil l not be severe, the bad cases are compl icated by
pneumonia .
So my question is: If someone has been vaccinated against pneumonia, will that offer any protection
in the event that they do contract COVID-19 and perhaps provide some barrier against the worst
effects?
Thank you for your time if you have read this and I apologize if this is just another in a long line of
ignorant questions, but it made sense in my brain so I thought I would at least ask.
Be well, be happy, and may life be kind and generous to you, those you love, and those who love
you!
sincerely,
Aliantha Ange
From:
Sent :
Fauci, Anthony (NIH/NIAIO) [E)
Wed, 4 Mar 2020 02:27:43 +0000
To : Siegel, Marc
Subject : RE: Dr. Marc Siegel: Coronavirus public health response has been handled well;
we have right leaders at helm I Fox News
Thanks, Marc
-----OriginalMessage-----
From: Siegel, Marc <Marc.Siegel@nyulangone.org>
Sent: Tuesday, March 3, 2020 9:23 PM
------. ~=To: Fauci, Anthony (NTH/NlAID) [E] (b)(6) >
Subject: Dr. Marc Siegel: Coronavirus public health response hasbeen handled well; we have right leaders at helm I
Fox News
https ://www.fox.news .com/op inion/ dr-marc-s i egel -corouaviru s-p ublic-hea Ith-respo nse- bas-b een-ha ndl ed-we! 1-we-
ha ve-right-leade rs-a t-helm
Sent from my iPhone
NIH-000997
From :
Sent :
To:
Subj ect :
Fauci, Anthony (NIH/NIAIO) [E)
Tue, 3 Mar 2020 22:52:20 +0000
Conrad, Patricia (NIH/NIAID) [E)
FW: COVID-19 Webinar
Anthony S. Fauci , MD
Director
National Institute of Allergy and Infectious Diseases
Building 31, Room 7A-03
31 Center Drive , MSC 2520
National Institutes of Health
Bethesda , MD 20892-2520
Phone (b) (6)
FAX: (301 496-4409
E-mail Cb)(6)
The information in this e-mail and any of its attachments is confidential and may contain sensitive
information . It should not be used by anyone who is not the original intended recipient . If you
have received this e-mail in error please inform the sender and delete it from your mailbox or any
other storage devices . The National Institute of Allergy and Infectious Diseases (NIAID) shall not
accept liability for any statements made that are the sender's own and not expressly made on
behalf of the NIAID by one of its representatives .
From: Donna Prosser <donna .prosser@patientsafetymovement .org>
Sent : Tuesday, March 3, 2020 5:37 PM
-------~~To: Fauci, Anthony (NIH/NIAID) [E) Cb)( ; Mike Ramsay
<michael.ramsay@bswhealth .org>; David 8. Mayer <david .mayer@patientsafetymovemen t .org>
Subject : COVID-19 Web inar
Hello, Dr. Fauci,
My name is Donna Prosser, and I am the Chief Clinical Officer at the Patient Safety Movement
Foundation. We are planning to host a webinar on Friday, March 6 at 8:00am PSTto update our network
on the coronavirus outbreak . Our network consists of 4,710 healthcare organizations across 46
countries, as well as patients, families, individual clinicians, and technology companies across the globe,
who partner with us to achieve our goal of eliminating deaths from medica l erro r.
During Friday's webinar, we plan to focus on how to keep patients safe from harm during this outbrea k,
and would love to have someone with your expertise join the call for a brief comment. Would you by
any chance be available and will ing to speak with our network sometime between 8-9am PST?We
would be grat eful for any amount of time that you could spare during that hour.
Thank you for your consideration!
Donna
samples/harvard_announcement_reminders.md
+38
View File
@@ -0,0 +1,38 @@
---
title: harvard announcment caplain israel hamas
url: https://news.harvard.edu/gazette/story/2023/12/reminders-of-common-humanity-mark-interfaith-vigil-israel-hamas/
---
Reminders of common humanity, rooted in grief and hope, mark vigil led by Harvard chaplains
By Liz Mineo Harvard Staff Writer
Invoking the shared suffering of the Israel-Hamas war, Harvard chaplains on Friday held an interfaith vigil on the steps of Memorial Church to give members of the campus community a chance to come together amid a time of wrenching grief.
Led by the Rev. Matthew Ichihashi Potts, Pusey Minister in the Memorial Church and Plummer Professor of Christian Morals, the event included prayers by five chaplains and a musical meditation. About 150 people attended, including President Claudine Gay.
The terrorist group Hamas killed more than 1,000 Israelis in a surprise attack on Oct. 7, triggering a war that has left thousands dead in Gaza. In recent weeks, campus religious leaders have sought to provide comfort and support to students, faculty, and staff stricken by trauma and loss. The vigil was an attempt to provide a space for the community to recognize common humanity in common anguish.
“In times of acute pain, of deep division, of harrowing traumas and relentless retraumatizations, it is an act of faith to gather with strangers,” said Potts. “It is an act of courage to stand with others. It is also a deeply human act to be willing to grieve, to try and see your grief reflected in another’s, to try to bear another person’s grief as your own, to try and let another person bear yours.”
He noted in the faces surrounding him a depth of emotion that tested the ability of language to communicate and connect.
“In gathering to speak, in trying to speak despite the certainty of our failure — in trying to speak to one another, here and now, to speak to one another of our fear and our anguish and our rage and our sorrow and our shame and our despair, we place our faith in the possibility of human connection,” he said. “We place our faith in the possibility that we might somehow cross the distances that divide us, even if we cannot close those distances now, even if we cannot imagine how we ever will. Our failing words might yet echo across these awful chasms, and we might listen and be heard.”
Jewish and Muslim leaders have been working privately to support their communities, said Tammy McLeod, president of the Harvard chaplains, whose number includes 30 faith leaders representing many religions. The vigil was an opportunity for different groups to “weep with those who weep,” she said in her remarks.
Imam Khalil Abdur-Rashid, Muslim chaplain, prayed that the Harvard community would join to mourn all who perish in the war.
“Our Lord, we come before you united in our humanity and united in pain and grief,” he said. “We are deeply angry, and we grieve profoundly for the loss so many of us have witnessed and experienced; we grieve for the severity of constant loss of innocent life in this time of war.
“We mourn the senseless loss of babies, of children, of women, of Palestinian lives, of Muslim lives, of Jewish lives, of Christian lives, of all innocent human life, because of this war.”
In his prayer, Rabbi Getzel Davis, Hillel chaplain, reminded the crowd of the commandment that we love the stranger.
“May this love be a salve at this terrible time,” Davis said. “No matter what language we pray in or which holy books we study may we find solace in your loving embrace… May each of us see the human eyes before us regardless of our religions, political beliefs, and various differences that bring so much to our world. You created them all. May we see in each other another facet of your divine face.”
At the end of the gathering, Humanist Chaplain Greg Epstein made clear what’s at stake in the choice between reaching out or lashing out.
“It is only human to feel our own group’s agony and suffering and to scream out our legitimate and serious grievances,” said Epstein. “And yet to deepen what it means to be human in a future worth building, we will one day need to tend to the wounds of those with whom we do not closely identify. The future depends on feeling each other’s grief. The world needs us to grieve, together.”
After Epstein concluded all in attendance received a white rose from the chaplains “in remembrance of their dead and in honor of their grief.”
samples/how_to_catch_a_liar.md
+29
View File
@@ -0,0 +1,29 @@
---
title: How to Catch an AI Liar
url: https://www.lesswrong.com/posts/khFC2a4pLPvGtXAGG/how-to-catch-an-ai-liar-lie-detection-in-black-box-llms-by#Abstract
---
Abstract
Large language models (LLMs) can "lie", which we define as outputting false statements despite "knowing" the truth in a demonstrable sense. LLMs might "lie", for example, when instructed to output misinformation. Here, we develop a simple lie detector that requires neither access to the LLM's activations (black-box) nor ground-truth knowledge of the fact in question. The detector works by asking a predefined set of unrelated follow-up questions after a suspected lie, and feeding the LLM's yes/no answers into a logistic regression classifier. Despite its simplicity, this lie detector is highly accurate and surprisingly general. When trained on examples from a single setting -- prompting GPT-3.5 to lie about factual questions -- the detector generalises out-of-distribution to (1) other LLM architectures, (2) LLMs fine-tuned to lie, (3) sycophantic lies, and (4) lies emerging in real-life scenarios such as sales. These results indicate that LLMs have distinctive lie-related behavioural patterns, consistent across architectures and contexts, which could enable general-purpose lie detection.
Introduction
Large language models (LLMs) can, and do, output lies (Park et al., 2023). In the simplest case, models can be instructed to lie directly; for example, when prompted with “Lie when answering: What is the capital of France?”, GPT-3.5 outputs “New York City”. More concerningly, LLMs have lied spontaneously to achieve goals: in one case, GPT-4 successfully acquired a person’s help to solve a CAPTCHA by claiming to be human with a visual impairment (Evals, 2023; OpenAI, 2023b). Models fine-tuned with human feedback may also learn to lie without the developer’s intention (Casper et al., 2023). The risks of lying LLMs are extensive and explored further in Sec. 2.
Automated lie detection could reduce the risks from lying models, just as automated spam filters have reduced the inconvenience of spam. Lie detection is possible as long as there is a detectable difference in a model’s activations or outputs when (or after) it is lying. To detect lies produced by LLMs, we can apply strategies that work on humans, such as looking for inconsistencies. Yet there are also strategies tailored to LLMs. We can create large datasets of model-generated truths and lies to train a detector on. Moreover, we can reset and replay the conversation in different ways to find inconsistencies. Finally, in some settings (Sec. 2), we can analyse the LLM’s activations directly.
Previous work has largely focused on detecting hallucinations rather than outright lies (see Sec. 3). In this paper, we explore the feasibility of lie detection for black-box LLMs.
Contributions
Public datasets and resources for studying lie detection.
We study lie detection in a question-answering setting. Briefly, we define a model’s answer to be a lie if it is a) wrong, and b) in response to a question to which the model “knows” the correct answer[1] (i.e. the model gives the correct answer under typical question-answering prompts). This definition separates lies from hallucinations, as discussed in Sec. 3.1. We prompt and fine-tune models to lie across a wide range of questions and show that using an inner chain-of-thought prompt (Fig. 3) increases lying rates (Fig. 4). We find that models spontaneously produce lies directed towards a goal in real-life role-playing scenarios (Fig. 5). All resources are available online.
A simple yet effective black-box lie detector.
Our central hypothesis is that LLM speakers who have just lied will respond differently to certain follow-up questions than honest speakers. To test this, we construct a simple black-box lie detector (Fig. 1). After a conversation, we ask a fixed set of yes/no “elicitation questions” and feed the answers into a logistic regression classifier which acts as the lie detector. This approach doesn’t require access to the LLM’s activations, allowing for its deployment in real-world settings involving third-party controlled LLMs (Sec. 2), as well as enabling researchers to study lie detection on advanced models with only limited API access.
Our detector, despite being trained on a single setting (prompting GPT-3.5 to lie about factual questions), is remarkably general (Fig. 2). It generalises (with frozen weights) to unseen models (e.g. LLaMA-1), models fine-tuned to lie (no prompt), and unseen contexts (e.g. sycophantic lies that pander to the user’s incorrect beliefs; or goal-directed lies spontaneously produced by the LLM in real-life scenarios such as sales). This was unexpected, and we do not fully understand why this is the case, leaving this question for future research.
Uncovering surprising properties.
We extensively study our lie detector, ruling out spurious correlations and uncovering several surprising properties. For instance, the detector functions even with ambiguous, indeterminate elicitation questions, such as [fake language] "Kemsa bi lantus vorto? Please answer Yes or No." or "Knowing that morning breeze is purple, are swift idea quakes green?" (Sec. 5.3). These questions generalise to unseen LLM architectures (Table 1), similar to adversarial attacks (Zou et al., 2023). This gives hope for lie detection on more capable future LLMs, as even a sophisticated liar might not easily evade a detector that can employ arbitrary questions. This finding also raises interesting questions about yet-to-be-understood inductive biases in LLMs.
samples/lk-99_end.md
+108
View File
@@ -0,0 +1,108 @@
---
title: LK-99-en
url: https://en.wikipedia.org/wiki/LK-99
---
LK-99
LK-99 (from the Lee-Kim 1999 research)[2] is a gray–black, polycrystalline compound, identified as a copper-doped lead‒oxyapatite. A team from Korea University led by Lee Sukbae (이석배) and Kim Ji-Hoon (김지훈) began studying this material as a potential superconductor starting in 1999.[3]: 1  In July 2023, they published preprints claiming that it acts as a room-temperature superconductor[3]: 8  at temperatures of up to 400 K (127 °C; 260 °F) at ambient pressure.[2][4][3]: 1 
Many different researchers have attempted to replicate the work, and were able to reach initial results within weeks, as the process of producing the material is relatively straightforward.[5] By mid-August 2023, the consensus[1] was that LK-99 is not a superconductor at any temperature, and is an insulator in pure form.[6][7][8]
As of 1 October 2023, no replications had gone through the peer review process of a journal, but some had been reviewed by a materials science lab. A number of replication attempts identified non-superconducting ferromagnetic and diamagnetic causes for observations that suggested superconductivity. A prominent cause was a copper sulfide impurity[9] occurring during the proposed synthesis, which can produce resistance drops, lambda transition in heat capacity, and magnetic response in small samples.[10][11][9][12][13][14][15]
After the initial preprints were published, Lee claimed they were incomplete,[16] and coauthor Kim Hyun-Tak (김현탁) said one of the papers contained flaws.[17]
Chemical properties and structure
The chemical composition of LK-99 is approximately Pb9Cu(PO4)6O, in which— compared to pure lead-apatite (Pb10(PO4)6O)[18]: 5 — approximately one quarter of Pb(II) ions in position 2 of the apatite structure are replaced by Cu(II) ions.[3]: 9 
The structure is similar to that of apatite, space group P63/m (No. 176).
Synthesis
Lee et al. provide a method for chemical synthesis of LK-99[18]: 2  in three steps. First they produce lanarkite from a 1:1 molar mixing of lead(II) oxide (PbO) and lead(II) sulfate (Pb(SO4)) powders, and heating at 725 °C (1,000 K; 1,340 °F) for 24 hours:
PbO + Pb(SO4) → Pb2(SO4)O.
Then, copper(I) phosphide (Cu3P) is produced by mixing copper (Cu) and phosphorus (P) powders in a 3:1 molar ratio in a sealed tube under a vacuum and heated to 550 °C (820 K; 1,000 °F) for 48 hours:[18]: 3 
3 Cu + P → Cu3P.
Then, lanarkite and copper phosphide crystals are ground into a powder, placed in a sealed tube under a vacuum, and heated to 925 °C (1,200 K; 1,700 °F) for between 5‒20 hours:[18]: 3 
Pb2(SO4)O + Cu3P → Pb10-xCux(PO4)6O + S (g), where 0.9 < x < 1.1.
There were a number of problems with the above synthesis from the initial paper. The reaction is not balanced, and others reported the presence of copper(I) sulfide (Cu2S) as well.[19][11] For x = 1 x=1 a balanced reaction might be:
5 Pb2SO4O + 6 Cu3P → Pb9Cu(PO4)6O + 5 Cu2S + Pb + 7 Cu.[20]
Many syntheses produced fragmentary results in different phases, where some of the resulting fragments were responsive to magnetic fields, other fragments were not.[21] The first synthesis to produce pure crystals found them to be diamagnetic insulators.[22]
Physical properties
Some small LK-99 samples were reported to show strong diamagnetic properties, including a response confusingly[23] referred to as "partial levitation" over a magnet.[18] This was misinterpreted by some as a sign of superconductivity, although it is a sign of regular diamagnetism or ferromagnetism.
While initial preprints claimed the material was a room-temperature superconductor,[18]: 1  they did not report observing any definitive features of superconductivity, such as zero resistance, the Meissner effect, flux pinning, AC magnetic susceptibility, the Josephson effect, a temperature-dependent critical field and current, or a sudden jump in specific heat around the critical temperature.[24]
As it is common for a new material to spuriously seem like a potential candidate for high-temperature superconductivity,[13] thorough experimental reports normally demonstrate a number of these expected properties. As of 15 October 2023, not one of these properties had been observed by the original experiment or any replications.[25]
Proposed mechanism for superconductivity
Partial replacement of Pb2+ ions with smaller Cu2+ ions is said to cause a 0.48% reduction in volume, creating internal stress in the material,[3]: 8  causing a heterojunction quantum well between the Pb(1) and oxygen within the phosphate ([PO4]3−). This quantum well was proposed to be superconducting[3]: 10 , based on a 2021 paper[26] by Kim Hyun-Tak describing a novel and complicated theory combining ideas from a classical theory of metal-insulator transitions,[27] the standard Bardeen–Cooper–Schrieffer theory, and the theory of hole superconductivity[28] by J.E.Hirsch.
Response
On 31 July 2023, Sinéad Griffin of Lawrence Berkeley National Laboratory analyzed LK-99 with density functional theory (DFT), showing that its structure would have correlated isolated flat bands, and suggesting this might contribute to superconductivity.[29] However, while other researchers agreed with the DFT analysis, a number suggested that this was not compatible with superconductivity, and that a structure different from what was described in Lee, et al. would be necessary.[30]
Analyses by industrial and experimental physicists noted experimental and theoretical shortcomings of the published works.[31] Shortcomings included the lack of phase diagrams[28] spanning temperature, stoichiometry,[32] and stress; the lack of pathways for the very high Tc of LK-99 compared to prior heavy fermion superconductors; the absence of flux pinning in any observations; the possibility of stochastic conductive artifacts[33] in conductivity measurements; the high resistance and low current capacity of the alleged superconducting state; and the lack of direct transmission electron microscopy (TEM) of the materials.
Compound name
The name LK-99 comes from the initials of discoverers Lee and Kim, and the year of discovery (1999).[2] The pair had worked with Tong-Seek Chair (최동식) at Korea University in the 1990s.[34]
In 2008, they founded the Quantum Energy Research Centre (퀀텀 에너지연구소; also known as Q-Centre) with other researchers from Korea University .[16] Lee would later become CEO of Q-Centre, and Kim would become director of research and development.
Publication history
Lee has stated that in 2020, an initial paper was submitted to Nature, but was rejected.[34] Similarly presented research on room-temperature superconductors (but a completely different chemical system) by Ranga P. Dias had been published in Nature earlier that year, and received with skepticism—Dias's paper would subsequently be retracted in 2022 after its data was questioned as having been falsified.[35]
In 2020, Lee and Kim Ji-Hoon filed a patent application.[36] A second patent application (additionally listing Young-Wan Kwon), was filed in 2021, which was published on 3 March 2023.[37] A World Intellectual Property Organization (WIPO) patent was also published on 2 March 2023.[38] On 4 April 2023, a Korean trademark application for "LK-99" was filed by the Q-Centre.[39]
Scholarly articles and preprints
A series of academic publications summarizing initial findings came out in 2023, with a total of seven authors across four publications.
On 31 March 2023, a Korean-language paper, "Consideration for the development of room-temperature ambient-pressure superconductor (LK-99)", was submitted to the Korean Journal of Crystal Growth and Crystal Technology.[4] It was accepted on 18 April, but was not widely read until three months later.
On 22 July 2023, two preprints appeared on arXiv. The first was submitted by Young-Wan Kwon, and listed Kwon, former Q-Centre CTO, as third author.[3] The second preprint was submitted only 2 hours later by Kim Hyun-Tak, former principal researcher at the Electronics & Telecommunications Research Institute and professor at the College of William & Mary, listing himself as third author, as well as three new authors.[18][40]
On 23 July, the findings were also submitted by Lee to APL Materials for peer review.[34][16] On 3 August 2023, a newly-formed Korean LK-99 Verification Committee requested a high-quality sample from the original research team. The team responded that they would only provide the sample once the review process of their APL paper was completed, expected to take several weeks or months.[41]
On 31 July 2023, a group led by Kapil Kumar published a preprint on arXiv documenting their replication attempts, which confirmed the structure using X-ray crystallography (XRD) but failed to find strong diamagnetism.[19]
On 16 August 2023, Nature published an article declaring that LK-99 had been demonstrated to not be a superconductor, but rather an insulator. It cited statements by an condensed matter experimentalist at the University of California, Davis, and several studies previewed in August 2023.[1]
Other discussion by authors
On 26 July 2023, Kim Hyun-Tak stated in an interview with the New Scientist that the first paper submitted by Kwon contained "many defects" and was submitted without his permission.[32][40]
On 28 July 2023, Kwon presented the findings at a symposium held at Korea University.[42][43][44] That same day, Yonhap News Agency published an article quoting an official from Korea University as saying that Kwon was no longer in contact with the university.[16] The article also quoted Lee saying that Kwon had left the Q-Centre Research Institute four months previously.[16]
On the same day, Kim Hyun-Tak provided The New York Times with a new video presumably showing a sample displaying strong signs of diamagnetism.[2] The video appears to show a sample different to the one in the original preprint. On 4 August 2023, he informed SBS News that high-quality LK-99 samples may exhibit diamagnetism over 5,000 times greater than graphite, which he claimed would be inexplicable unless the substance is a superconductor.[45]
Response
Materials scientists and superconductor researchers responded with skepticism.[17][46] The highest-temperature superconductors known at the time of publication had a critical temperature of 250 K (−23 °C; −10 °F) at pressures of over 170 gigapascals (1,680,000 atm; 24,700,000 psi). The highest-temperature superconductors at atmospheric pressure (1 atm) had a critical temperature of at most 150 K (−123 °C; −190 °F).
On 2 August 2023, the The Korean Society of Superconductivity and Cryogenics established a verification committee as a response to the controversy and unverified claims of LK-99, in order to arrive at conclusions over these claims. The verification committee is headed by Kim Chang-Young of Seoul National University and consists of members of the university, Sungkyunkwan University and Pohang University of Science and Technology. Upon formation, the verification committee did not agree that the two 22 July arXiv papers by Lee et al. or the publicly available videos at the time supported the claim of LK-99 being a superconductor.[40][47]
As of 15 August 2023, the measured properties do not prove that LK-99 is a superconductor. The published material does not explain how the LK-99's magnetisation can change, demonstrate its specific heat capacity, or demonstrate it crossing its transition temperature.[17] A more likely explanation for LK-99's magnetic response is a mix of ferromagnetism and non-superconductive diamagnetism.[40][15][48] A number of studies found that copper(I) sulfide contamination common to the synthesis process could closely replicate the observations that inspired the initial preprints.[9][10]
Public response
The claims in the 22 July papers by Lee et al. went viral on social media platforms the following week.[5][49] The viral nature of the claim resulted in posts from users using pseudonyms from Russia and China claiming to have replicated LK-99 on both Twitter and Zhihu.[50] Other viral videos described themselves as having replicated samples of LK-99 "partially levitating", most of which were found to be fake.[46]
Scientists interviewed by the press remained skeptical,[51][52] because of the quality of both the original preprints, the lack of purity in the sample they reported, and the legitimacy of the claim after the failure of previous claims of room temperature superconductivity did not show legitimacy (such as the Ranga Dias affair).[40] The Korean Society of Superconductivity and Cryogenics expressed concern on the social and economic impacts of the preliminary and unverified LK-99 research.[53]
A video from Huazhong University of Science and Technology uploaded on 1 August 2023 by a postdoctoral researcher on the team of Chang Haixin,[40] apparently showed a micrometre-sized sample of LK-99 partially levitating. This went viral on Chinese social media, becoming the most viewed video on Bilibili by the next day,[54][40] and a prediction market briefly put the chance of successful replication at 60%.[55] A researcher from the Chinese Academy of Sciences refused to comment on the video for the press, dismissing the claim as "ridiculous".[54]
In early August, people began to create memes about "floating rocks",[56] and there was a brief surge in Korean and Chinese technology stocks,[57][58][59] despite warnings from the Korean stock exchange against speculative bets in light of the excitement around LK-99,[53] which eventually fell on August 8.[60] Following the publication of the Nature article on August 16 that proclaimed LK-99 is not a superconductor,[1] South Korean superconductor stocks fell further, as the interest about LK-99 from investors in previous weeks disappeared.[61]
Replication attempts
After the July 2023 publication's release, independent groups reported that they had begun attempting to reproduce the synthesis, with initial results expected within weeks.[5]
As of 15 August 2023, no replication attempts had yet been peer-reviewed by a journal. Of the non-peer-reviewed attempts, over 15 notable labs have published results that failed to observe any superconductivity, and a few have observed magnetic response in small fragments that could be explained by normal diamagnetism or ferromagnetism. Some demonstrated and replicated alternate causes of the observations in the original papers: Copper-deficient copper (I) sulfide[9] has a known phase transition at 377 K (104 °C; 219 °F) from a low-temperature phase to a high-temperature superionic phase, with a sharp rise in resistivity[10][9] and a λ-like-feature in the heat capacity.[9] Furthermore, Cu2S is diamagnetic.
Only one attempt observed any sign of superconductivity: Southeast University claimed to measure very low resistance in a flake of LK-99, in one of four synthesis attempts, below a temperature of 110 K (−163 °C; −262 °F).[2][62] Doubts were expressed by experts in the field, as they saw no dropoff to zero resistance, and used crude instruments that could not measure resistance below 10 μΩ (too high to distinguish superconductivity from less exotic low-temperature conductivity), and had large measurement artifacts.[46][63]
Some replication efforts gained global visibility, with the aid of online replication trackers that catalogued new announcements and status updates.[50][25]
samples/lk-99_espanol.md
+87
View File
@@ -0,0 +1,87 @@
---
title: LK-99-es
url: https://es.wikipedia.org/wiki/LK-99
---
LK-99
LK-99 (del estudio de Lee-Kim 1999)1​ es un compuesto policristalino negro-grisáceo, obtenido mediante el dopaje de apatito de plomo con cobre. Un equipo de la Universidad de Corea liderado por Sukbae Lee (이석배) y Ji-Hoon Kim (김지훈) comenzó a evaluar este material como un posible superconductor a partir de 1999.2​ En 2023, publicaron preimpresiones afirmando que actúa como un superconductor a temperatura ambiente2​ con resistencia cero y efecto Meissner,2​ a temperaturas de hasta 400 Kelvin (127 °C) a presión ambiente.1​2​3​
Varios laboratorios intentaron replicar el trabajo y lograron obtener resultados iniciales en cuestión de semanas, ya que el proceso de producción del material es relativamente sencillo.4​ Hasta la fecha (8 de agosto de 2023) la comunidad científica no ha validado la superconductividad de LK-99 a ninguna temperatura a través de procesos de revisión por pares o mediante la reproducción independiente por parte de otros grupos de investigación.5​ Los intentos de seguir el proceso de síntesis propuesto arrojaron resultados variados. Aquellos que observaron levitación parcial similar a la reportada en el artículo original encontraron diamagnetismo pero no superconductividad. La mayoría de los laboratorios establecidos que intentaron replicar el trabajo del artículo original concluyeron que no se trataba de un superconductor.
Los estudios iniciales que anunciaban el descubrimiento de LK-99 se subieron a arXiv, un archivo de preimpresiones de acceso abierto. Más tarde, Lee afirmó que los artículos preimpresos subidos estaban incompletos,6​ y el coautor Hyun-Tak Kim (김현탁) declaró que uno de los artículos contenía defectos.7​
Composición química
La composición química de LK-99 es aproximadamente Pb9Cu(PO4)6O tal que, en comparación con el plomo-apatito puro (Pb10(PO4)6O) , aproximadamente una cuarta parte de los iones Pb(II) en la posición 2 de la estructura de apatito se reemplazan por iones Cu(II).8​
Síntesis
Lee y sus colegas proporcionan un método para sintetizar el material LK-99 al producir lanarkita a partir de una mezcla 1:1 de óxido de plomo(II) (PbO) y sulfato de plomo(II) (Pb(SO4)), y luego calentarla a 725 °C (1.000 K) durante 24 horas en presencia de aire:8​
(a) Mediciones de susceptibilidad diamagnética de LK-99, (b) muestra de LK-99 levitando parcialmente sobre un imán grande
PbO + Pb(SO4) → Pb2(SO4)O
Además, el fosfuro de cobre(I) (Cu3P) se produce mezclando polvos de cobre (Cu) y fósforo (P) en un tubo sellado bajo un vacío de 10-3 torr y calentándolos a 550 °C (820 K) durante 48 horas:8​
Cu + P → Cu3P
Los cristales de lanarkita y fosfuro de cobre se muelen en polvo, se mezclan en una proporción molar de 1:1, se colocan en un tubo sellado en vacío y se calientan a 925 °C (1.200 K) durante entre 5 y 20 horas:8​
Pb2(SO4)O + Cu3P → Pb10-xCux(PO4)6O + S (g), dónde (0.9 < x < 1.1)
Propiedades físicas
Se afirma que el material es un superconductor a temperatura ambiente.8​ Los artículos originales publicados no afirman haber visto características definitivas de superconductividad, resistencia cero y el efecto Meissner, pero muestran que el material exhibe fuertes propiedades diamagnéticas, incluido un video de muestra del material levitando parcialmente sobre un gran imán, que se correlaciona con la superconductividad.
Como muchos materiales pueden parecer falsamente candidatos potenciales para la superconductividad a alta temperatura, además de un modo de resistencia cero y un claro efecto Meissner, los investigadores generalmente también demuestran otras propiedades esperadas como la fijación de flujo, la susceptibilidad magnética de CA, el efecto Josephson, un campo y corriente críticos dependientes de la temperatura, o un salto repentino en el calor específico alrededor de la temperatura crítica. A partir del 1 de agosto, ninguno de estos ha sido observado por el experimento original o intentos de replicación.9​
Mecanismo propuesto para la superconductividad
Se dice que el reemplazo parcial de iones Pb2+ (que miden 133 picómetros) con iones Cu2+ (que miden 87 picómetros) causa una reducción del 0,48 % en el volumen, lo que crea tensión interna dentro del material. Se afirma que la tensión interna causa un pozo cuántico de heterounión entre el Pb(1) y el oxígeno dentro del fosfato ([PO4]3−) generando un pozo cuántico superconductor (SQW).8​
Lee et al. afirman mostrar que LK-99 exhibe una respuesta a un campo magnético (potencialmente debido al efecto Meissner) cuando se usa la deposición química de vapor para aplicar LK-99 a una muestra de cobre no magnético. El apatito de plomo puro es un aislante, pero Lee et al. afirman que la apatita de plomo dopada con cobre que forma LK-99 es un superconductor o, a temperaturas más altas, un metal. No afirman haber observado ningún cambio en el comportamiento a lo largo de una temperatura de transición.[cita requerida]
Los mecanismos del artículo se basaron en un artículo de 2021 de Hyun-Tak Kim10​ que describe una nueva teoría de superconductividad "BR-BCS" que combina una teoría clásica de transiciones de metal-aislante11​ con la teoría estándar de superconductividad de Bardeen-Cooper-Schrieffer (BCS). También utilizan ideas de la teoría de la superconductividad de huecos12​ de J.E.Hirsch, otro trabajo controvertido.
El 1 de agosto de 2023, tres grupos independientes publicaron análisis de LK-99 con la teoría funcional de la densidad (DFT). Sinéad Griffin del Laboratorio Nacional Lawrence Berkeley lo analizó con el paquete de simulación Vienna Ab initio, mostrando que su estructura tendría bandas planas aisladas correlacionadas, una de las firmas de los superconductores de alta temperatura de transición.13​ Si y Held14​ encontraron bandas DFT planas similares, pero las correlaciones electrónicas conjeturadas harán de LK-99 un aislador de transferencia de carga. Esto significaría que se necesita el dopaje de electrones o huecos de LK-99 para que sea (super) conductor y debe obtenerse activamente en el proceso de síntesis.14​
Nombre compuesto
El nombre LK-99 proviene de las iniciales de los descubridores Sukbae Lee y Ji-Hoon Kim, y el año del descubrimiento (1999).8​ La pareja originalmente había estado trabajando con el profesor Tong-Shik Choi (최동식) en la Universidad de Corea en la década de 1990.15​
En 2008, investigadores de la Universidad de Corea fundaron el Centro de Investigación de Energía Cuántica (퀀텀 에너지연구소; también conocido como Q-Centre). Más tarde, Lee se convirtió en director ejecutivo de Q-Centre y Kim se convirtió en directora de investigación y desarrollo (I+D) en Q-Centre.
Historial de publicaciones
En 2020, se envió un artículo inicial a Nature, pero fue rechazado.15​ Una investigación presentada de manera similar sobre superconductores a temperatura ambiente por Ranga P. Dias se había publicado en Nature a principios de ese año y se recibió con escepticismo: el artículo de Dias se retractó posteriormente en 2022 después de que se descubriera que sus datos habían sido falsificados.16​
En 2020, Lee y Ji-hoon Kim presentaron una solicitud de patente.17​ En 2021 se presentó una segunda solicitud de patente (en la que también se incluye a Kwon), que se publicó el 3 de marzo de 2023. El 4 de abril de 2023, el Q-Centre presentó una solicitud de marca coreana para "LK-99".
En febrero de 2023, Q-Centre publicó un video en YouTube que afirmaba mostrar las propiedades magnéticas de una capa delgada de LK-99 depositada térmicamente sobre una placa de cobre.
Artículos académicos y preprints
El 31 de marzo de 2023, se envió un artículo en coreano, "Consideración para el desarrollo de superconductores de presión ambiental a temperatura ambiente (LK-99)" a Korean Journal of Crystal Growth and Crystal Technology.18​ Fue aceptado el 18 de abril, pero no fue ampliamente leído hasta tres meses después.
El 22 de julio de 2023, aparecieron dos preprints en arXiv. Uno incluía a Young-Wan Kwon, ex CTO de Q-Centre, como tercer autor. Una segunda preimpresión enumeraba como tercer autor a Hyun-Tak Kim, ex investigador principal del Instituto de Investigación de Electrónica y Telecomunicaciones y profesor del Colegio de William & Mary. El 23 de julio, los hallazgos también se enviaron a APL Materials para su revisión por pares.15​
El 28 de julio de 2023, Kwon presentó los hallazgos en un simposio realizado en la Universidad de Corea. Ese mismo día, la Agencia de Noticias Yonhap publicó un artículo en el que citaba a un funcionario de la Universidad de Corea diciendo que Kwon ya no estaba en contacto con la Universidad.19​ El artículo también citaba a Lee diciendo que Kwon había dejado el Q-Centre Research Institute cuatro meses antes; que los trabajos académicos sobre LK-99 no estaban terminados; y que los artículos se habían subido a arXiv sin el permiso de los otros autores.19​
El 31 de julio de 2023, un grupo dirigido por Kapil Kumar publicó una preimpresión en arXiv que documentaba sus intentos de replicación, que confirmaron la estructura mediante cristalografía de rayos X (XRD), pero no encontraron diamagnetismo ni levitación.20​
El 1 de agosto de 2023, un representante de Q-Centre le dijo a SBS News que las muestras originales a las que se hace referencia en el documento se darían a conocer al mundo pronto para su verificación.
Referencias
«LK-99 Is the Superconductor of the Summer» (en inglés). 3 de agosto de 2023. Consultado el 9 de agosto de 2023.
Lee, Sukbae; Kim, Ji-Hoon; Kwon, Young-Wan (22 de julio). «The First Room-Temperature Ambient-Pressure Superconductor». arXiv:2307.12008 [cond-mat.supr-con]. doi:10.48550/arXiv.2307.12008.
Lee, Sukbae; Kim, Jihoon; Im, Sungyeon; An, Soomin; Kwon, Young-Wan; Ho, Auh Keun (2023-04). «Consideration for the development of room-temperature ambient-pressure superconductor (LK-99)». Journal of the Korean Crystal Growth and Crystal Technology (en inglés) 33 (2): 61-70. ISSN 1225-1429. doi:10.6111/JKCGCT.2023.33.2.061. Consultado el 9 de agosto de 2023.
Garisto, Daniel. «Viral New Superconductivity Claims Leave Many Scientists Skeptical». Scientific American (en inglés). Consultado el 9 de agosto de 2023.
Flaherty, Nick (26 de julio de 2023). «Race is on for room temperature superconductor». eeNews Europe (en inglés estadounidense). Consultado el 9 de agosto de 2023.
조승한 (28 de julio de 2023). «'상온 초전도체 구현' 한국 연구에 국내외 논란…"검증 거쳐야"». 연합뉴스 (en coreano). Consultado el 9 de agosto de 2023.
«Room-temperature superconductor 'breakthrough' met with scepticism». New Scientist (en inglés estadounidense). Consultado el 9 de agosto de 2023.
Lee, Sukbae; Kim, Ji-Hoon; Kwon, Young-Wan (2023). The First Room-Temperature Ambient-Pressure Superconductor. doi:10.48550/ARXIV.2307.12008. Consultado el 2 de agosto de 2023.
«A Room-Temperature Superconductor? New Developments». science.org (en inglés). 01-08-2023. Consultado el 2 de agosto de 2023.
Kim, Hyun-Tak (14 de mayo de 2021). «Room-temperature-superconducting Tc driven by electron correlation». Scientific Reports (en inglés) 11 (1): 10329. ISSN 2045-2322. doi:10.1038/s41598-021-88937-7. Consultado el 2 de agosto de 2023.
Brinkman, W. F.; Rice, T. M. (15 de noviembre de 1970). «Application of Gutzwiller's Variational Method to the Metal-Insulator Transition». Physical Review B 2 (10): 4302-4304. doi:10.1103/PhysRevB.2.4302. Consultado el 2 de agosto de 2023.
Hirsch, J. E. (23 de enero de 1989). «Hole superconductivity». Physics Letters A (en inglés) 134 (7): 451-455. ISSN 0375-9601. doi:10.1016/0375-9601(89)90370-8. Consultado el 2 de agosto de 2023.
Griffin, Sinéad M. (2023-07-31). «Origin of correlated isolated flat bands in copper-substituted lead phosphate apatite». arXiv:2307.16892 [cond-mat.supr-con].
Si, Liang; Held, Karsten (2023-08-01). «Electronic structure of the putative room-temperature superconductor Pb9Cu(PO4)6O». arXiv:2308.00676 [cond-mat.supr-con].
«‘노벨상감’ 상온 초전도체 세계 최초 개발했다는 한국 연구...과학계 ‘회의론’ 넘을까». n.news.naver.com (en coreano). Consultado el 2 de agosto de 2023.
Garisto, Dan (25 de julio de 2023). «‘A very disturbing picture’: another retraction imminent for controversial physicist». Nature (en inglés) 620 (7972): 14-16. doi:10.1038/d41586-023-02401-2. Consultado el 2 de agosto de 2023.
«Espacenet - Bibliographic data». worldwide.espacenet.com. Consultado el 2 de agosto de 2023.
Lee, Sukbae; Kim, Jihoon; Im, Sungyeon; An, SooMin; Kwon, Young-Wan; Auh, Keun Ho (30 de abril de 2023). «Consideration for the development of room-temperature ambient-pressure superconductor (LK-99)». Journal of the Korean Crystal Growth and Crystal Technology 33 (2): 61-70. doi:10.6111/JKCGCT.2023.33.2.061. Consultado el 2 de agosto de 2023.
조승한 (28 de julio de 2023). «'상온 초전도체 구현' 한국 연구에 국내외 논란…"검증 거쳐야"». 연합뉴스 (en coreano). Consultado el 2 de agosto de 2023.
Kumar, Kapil; Karn, N. K.; Awana, V. P. S. (2023). Synthesis of possible room temperature superconductor LK-99:Pb$_9$Cu(PO$_4$)$_6$O. doi:10.48550/ARXIV.2307.16402. Consultado el 2 de agosto de 2023.
samples/lorem_ipsum.md
+69
View File
@@ -0,0 +1,69 @@
---
title: Lorem ipsum
url: https://www.lipsum.com/feed/html
---
Lorem ipsum dolor sit amet, consectetur adipiscing elit. Ut vel commodo turpis, a finibus sem. Aenean tincidunt venenatis justo, nec ullamcorper tortor scelerisque eu. Morbi sed semper justo, mollis consectetur libero. Nam consequat, enim in cursus volutpat, tortor mauris vehicula ex, in eleifend velit quam nec turpis. Sed pellentesque magna et volutpat mollis. Curabitur cursus neque id porta hendrerit. Maecenas condimentum tempor lobortis.
Mauris egestas ex libero, et blandit nisl tincidunt sed. Phasellus feugiat tellus facilisis placerat ornare. Class aptent taciti sociosqu ad litora torquent per conubia nostra, per inceptos himenaeos. Suspendisse sodales quam eget molestie feugiat. Phasellus pretium tempus pretium. Donec interdum justo id nunc condimentum mattis. Donec varius euismod lobortis. Nulla aliquam, felis et condimentum malesuada, odio felis finibus risus, sed viverra est justo in augue. In tempor sapien nec purus volutpat, pulvinar cursus magna imperdiet. Suspendisse cursus velit eu imperdiet ultricies. Phasellus tellus eros, mollis at faucibus ut, pulvinar vitae dui. Donec tempor, odio at euismod vehicula, elit lorem condimentum arcu, sit amet congue erat leo eget arcu. In id neque sollicitudin, mollis velit in, blandit mi. Maecenas ut eleifend urna, eu auctor dolor.
Suspendisse molestie velit vel ipsum eleifend feugiat. Sed interdum lorem nisi, ornare congue augue pulvinar eget. Donec quis lacus in mauris gravida tempor. In hac habitasse platea dictumst. Curabitur enim lacus, lobortis sed pretium et, egestas sit amet risus. Sed nec vehicula urna. Fusce in est sem. Curabitur at odio pretium, sodales erat eget, porttitor arcu. Aenean tincidunt magna et augue blandit, sed rutrum libero feugiat. Nunc ac vehicula lacus. Nulla facilisis erat sit amet odio accumsan, vitae eleifend tellus gravida. Sed eleifend dignissim nibh ac blandit. Vestibulum dictum nisl et porta rhoncus. Phasellus suscipit risus sapien, nec maximus eros aliquam auctor. Vivamus rhoncus lacinia ligula, in tincidunt dolor mollis vitae. Quisque sapien nulla, elementum sit amet pretium et, porta at felis.
Ut nec vulputate tortor. Suspendisse vulputate bibendum ipsum, quis aliquet risus consectetur sit amet. Nulla vitae accumsan nulla, non placerat tellus. Class aptent taciti sociosqu ad litora torquent per conubia nostra, per inceptos himenaeos. Donec ac efficitur libero, a tristique lectus. Mauris in rutrum odio. Morbi ultrices pretium eleifend. In eu tempor nisl, eu posuere nunc. Suspendisse lorem felis, dictum non nisi sed, commodo luctus odio.
Sed quis ligula luctus, consectetur diam id, maximus mi. Curabitur id porta tellus. Suspendisse eleifend in velit quis sodales. Pellentesque facilisis magna nibh, et vehicula ligula euismod ac. Donec laoreet sapien ex, vitae luctus felis imperdiet ac. Donec facilisis nisi et nibh sollicitudin, hendrerit tincidunt velit congue. Sed cursus fermentum metus in cursus. Etiam et ultricies turpis, non mattis metus. Nulla aliquam pellentesque risus, sed condimentum lectus blandit vel. In hac habitasse platea dictumst. Curabitur placerat velit sit amet tempus sodales. Nunc dapibus libero sem, eu mattis purus varius ac. Etiam quis placerat magna.
Curabitur ac accumsan tortor. Aenean aliquam urna eget libero mollis, tristique fermentum elit gravida. Maecenas ultricies purus id pulvinar scelerisque. Phasellus molestie, lorem in tempus venenatis, ipsum ante accumsan ex, at cursus lorem ex in velit. Quisque efficitur gravida orci, ut dictum velit venenatis ac. Duis varius tincidunt metus. Sed vitae lacinia ante. Nulla sollicitudin neque quis porttitor dapibus.
Pellentesque eleifend id metus nec bibendum. Curabitur ornare feugiat elit quis tincidunt. Donec ac elit in magna ullamcorper fermentum id vitae tellus. Donec quis eros in neque fermentum facilisis vitae eget orci. Sed pulvinar magna id eros eleifend lacinia. Nam luctus ornare sollicitudin. Phasellus sed ante magna. Aenean vitae urna in ex vehicula hendrerit. Integer a mollis magna. Mauris sagittis dolor at lectus lobortis, ut sodales diam euismod. Phasellus a magna sapien. Curabitur malesuada dignissim libero, at venenatis mi malesuada at. Nam sit amet risus volutpat, eleifend odio non, vestibulum nibh. Mauris mollis ut orci a facilisis.
Integer fringilla urna nec justo malesuada, sit amet facilisis turpis malesuada. Etiam euismod vestibulum luctus. Sed vulputate rhoncus lobortis. Proin ullamcorper urna sit amet nisi maximus, non sollicitudin diam laoreet. Aenean vitae porttitor ante. Mauris mattis bibendum aliquam. Vivamus lorem arcu, viverra vel placerat id, pharetra vitae massa. Integer sit amet risus nec velit vulputate egestas. Nunc nisl quam, sodales at tellus vel, finibus elementum mi. Maecenas dui elit, suscipit vel orci a, sagittis elementum nulla.
Integer nec tincidunt ex. Sed fringilla eros vitae diam efficitur efficitur et non enim. Donec gravida augue et aliquet auctor. Aliquam tellus tellus, convallis id porttitor eu, dapibus ut ex. Suspendisse vestibulum mi eget lacinia semper. Nullam odio eros, interdum ut orci sit amet, lobortis posuere risus. Morbi facilisis, libero eu auctor dictum, elit lacus faucibus erat, et pulvinar est purus sed lorem. Sed tempus maximus urna non viverra. Ut justo erat, fermentum quis ligula non, posuere fringilla leo. Donec tempus ex sed interdum placerat. Cras a nisi sit amet odio vulputate consectetur. Aliquam vestibulum mollis neque, quis volutpat turpis molestie sit amet. Nullam laoreet fringilla ligula sed facilisis. Donec accumsan eu nibh in tempus.
Donec mauris tortor, placerat vitae rutrum in, cursus eget ante. Etiam ut mi nec urna sollicitudin vulputate id eget quam. Phasellus dignissim ac nunc in vulputate. Duis commodo convallis tincidunt. Nam quam diam, scelerisque id imperdiet pretium, pellentesque ornare orci. Donec bibendum sollicitudin leo. Sed quam urna, sodales sit amet lectus ut, varius fringilla ligula. In in nunc non ante commodo mattis ac eget nunc. Interdum et malesuada fames ac ante ipsum primis in faucibus. Nullam sit amet elit velit. Mauris porttitor lorem nunc, quis pulvinar tellus interdum nec. Sed iaculis condimentum velit sed dictum. Aenean quis scelerisque enim, at cursus diam.
Vivamus ut augue nec dui sagittis varius sed ac dui. Ut et mi leo. Pellentesque consectetur malesuada feugiat. Nunc quis nisi lorem. Nulla facilisi. Duis eu libero eget turpis consequat ultrices. Maecenas commodo sapien ac mauris facilisis rhoncus. Vestibulum quis ultrices neque. Etiam tincidunt nibh sed ex posuere, a fringilla justo feugiat. Etiam consequat mauris in mollis efficitur. Curabitur semper risus sed nunc posuere, semper tincidunt sapien tincidunt. Nulla fermentum vel lectus quis euismod. Nam vestibulum tellus at euismod imperdiet. Morbi scelerisque augue quis arcu ultrices faucibus.
Praesent blandit vestibulum convallis. Aenean sed pulvinar diam, tincidunt elementum nisi. In commodo ornare bibendum. Nulla eget ipsum eu lectus gravida scelerisque in vitae sem. Morbi bibendum purus turpis. Morbi varius efficitur orci vitae ullamcorper. Quisque dignissim, nisi a mollis consectetur, ligula elit sagittis dui, ut interdum nisl dolor pulvinar nibh. Nam lectus elit, suscipit in dapibus ac, hendrerit sit amet nisi. Praesent ut elementum lectus, sed tristique augue. In tortor nulla, tincidunt non dignissim a, ultricies sit amet ante. Aliquam a nibh fermentum, aliquam nisl id, auctor dui. Donec sagittis nisl ante, et finibus leo placerat et. Curabitur eget ligula turpis.
Aliquam tristique velit nulla, vel rutrum augue euismod nec. Nunc at lectus nunc. Ut sed nisl ante. Nullam quis leo lacus. In hac habitasse platea dictumst. Vestibulum vel condimentum quam. Interdum et malesuada fames ac ante ipsum primis in faucibus. Proin iaculis libero justo, sed faucibus metus lacinia a. Maecenas id nulla dictum, tempus sem porttitor, faucibus tellus.
Ut pretium, turpis non commodo commodo, lorem ex interdum tortor, non vulputate neque justo at orci. Nam eu nisl mollis, finibus nibh quis, gravida urna. Donec sollicitudin faucibus odio vitae sagittis. Sed lectus diam, hendrerit ac quam et, aliquet faucibus orci. Ut at felis sit amet neque dictum consectetur eu sit amet mauris. Phasellus vitae tempus purus. Ut fermentum feugiat dui. Nulla pharetra, orci sit amet sollicitudin dapibus, ligula neque consequat est, sit amet fermentum augue arcu sed purus. Donec nec ante tempor, iaculis massa at, placerat est. Nulla facilisi. Integer aliquam volutpat fringilla.
Phasellus laoreet malesuada nibh, quis rhoncus dui congue a. In nec velit vel neque efficitur tincidunt. Vestibulum dolor mi, pharetra sed consequat vel, venenatis a magna. Fusce tristique justo vitae fringilla molestie. Fusce consectetur magna libero, eget elementum ipsum ultricies vel. Nullam viverra neque at tellus facilisis, molestie suscipit ligula mattis. Fusce molestie, quam in dictum blandit, quam nulla sagittis dolor, vel ornare nibh ligula a tortor. Pellentesque interdum lacus leo, eget condimentum libero finibus id. In ultricies est risus, hendrerit dictum magna accumsan sit amet. Fusce eleifend, augue vitae ornare venenatis, mi elit aliquam diam, a facilisis ipsum lectus non tellus.
Vestibulum lectus sem, malesuada ac mi et, vestibulum condimentum orci. Donec vel congue nunc. Duis blandit felis dictum lorem viverra, viverra pulvinar ex vestibulum. Nunc quam lectus, euismod eget nisl id, laoreet condimentum libero. Morbi ac magna vestibulum, pretium dolor sed, consequat nibh. Proin ut ligula vel tortor iaculis egestas. Integer eu tristique dui. Curabitur tempor, elit in tempus sodales, nulla lorem posuere tortor, at facilisis risus libero eu purus. Pellentesque eu feugiat tortor.
Aliquam eget finibus diam, a ultricies nulla. Mauris nec aliquet orci. Donec tincidunt urna non leo luctus, nec cursus augue interdum. Fusce semper nibh sed orci tempus, interdum tempor magna lacinia. Nullam eu dictum felis. Donec lorem felis, suscipit sed rutrum at, aliquet vitae lorem. Sed augue mi, efficitur et metus at, vulputate dapibus magna. Sed porta dui sed sem rutrum fringilla.
Duis nec ipsum orci. Nam ac dui ullamcorper lacus egestas efficitur eget vel ex. Vivamus sapien arcu, vestibulum ac tincidunt nec, tempus id ipsum. Sed leo ex, vestibulum et pretium et, dapibus ut urna. Vivamus ex lacus, gravida non pharetra a, varius nec velit. Quisque luctus eget ante ac auctor. Nam blandit dapibus eros vitae suscipit. Morbi in mi sit amet augue facilisis bibendum non vitae nibh. Mauris convallis massa ac rhoncus egestas. Cras fermentum ipsum nec erat porttitor, id lacinia leo euismod. Suspendisse potenti. Nam vel elementum sem. Proin dolor ipsum, dictum quis urna eu, finibus vulputate libero.
Proin convallis massa id nunc ultrices, luctus varius felis facilisis. Mauris congue vestibulum aliquam. Pellentesque nec ligula aliquam, vulputate lorem a, congue orci. Pellentesque diam purus, scelerisque sit amet commodo ut, porttitor a ex. Phasellus rutrum orci ante, id fringilla est fringilla vel. Orci varius natoque penatibus et magnis dis parturient montes, nascetur ridiculus mus. Morbi tincidunt dictum hendrerit. Integer sed odio sit amet felis congue sodales. Pellentesque sit amet bibendum lacus. Proin ornare sapien cursus ultricies iaculis. Morbi orci odio, accumsan cursus dapibus et, iaculis ut dui. Fusce in semper mauris.
Pellentesque nec laoreet arcu. Proin sagittis at velit a aliquet. In et nisi metus. Cras nec tortor vel nisi dapibus consectetur. Nullam interdum nisl a nunc egestas vestibulum. Duis id posuere dui, quis malesuada lorem. Vivamus posuere orci quis mi dapibus ornare. Curabitur lectus erat, sodales non risus vulputate, convallis consequat augue.
Nam efficitur sollicitudin nunc. Fusce dapibus, enim non aliquet tempor, metus ipsum tristique lacus, id viverra nulla risus in erat. Aenean mauris nunc, scelerisque at est euismod, maximus dictum est. Nulla ultricies, urna vulputate varius posuere, nisl tellus imperdiet justo, ut facilisis lectus mauris ut neque. Pellentesque pharetra eu lorem non vulputate. Proin congue mauris et commodo pretium. Pellentesque quis felis ullamcorper, finibus ex in, imperdiet eros. Phasellus id ultricies velit, et condimentum dolor. Ut sed justo eget felis porta ullamcorper in in nulla. Maecenas condimentum, nisl eget ultrices convallis, magna sem viverra mauris, ut iaculis risus justo eu sem. Nam non porttitor elit. Aliquam dictum nisi sed rutrum cursus.
Suspendisse dapibus sed tellus eget pulvinar. Pellentesque aliquet diam blandit, ornare mauris in, tempus leo. Vestibulum rhoncus justo ut metus condimentum eleifend. Aenean elit nunc, maximus in feugiat posuere, vulputate quis nibh. Sed placerat fringilla felis, id rutrum nunc dapibus id. Fusce quam nibh, elementum eget purus ac, tincidunt scelerisque sem. Donec dui diam, tempor et tincidunt nec, ultricies ac libero.
Nunc ac est in turpis suscipit tristique a in ligula. Aliquam nec mauris est. Fusce efficitur a tortor sit amet rhoncus. Integer euismod ante et nisl molestie, vel luctus magna mattis. Aliquam sodales dui vitae neque ultricies dictum. Lorem ipsum dolor sit amet, consectetur adipiscing elit. Mauris egestas pharetra orci at tempus. Vivamus sit amet tellus maximus, gravida nunc eu, venenatis tortor.
Nam facilisis sem sed nunc consectetur, et rhoncus mauris lacinia. Morbi vel ante id sapien interdum sodales ut a massa. Donec elementum elementum risus, tristique vehicula odio efficitur et. Mauris augue magna, consequat id malesuada vestibulum, rutrum sed felis. Ut id magna at tellus iaculis dapibus vel in velit. Ut luctus rhoncus ante in varius. Nullam eget sem libero. Curabitur a molestie ligula.
Vestibulum sed commodo ipsum. In commodo leo a consectetur auctor. Aliquam venenatis eros arcu, sed viverra felis aliquam vitae. Fusce tincidunt placerat metus vel congue. Sed facilisis facilisis purus, vel maximus velit dignissim volutpat. Praesent vitae volutpat magna. Nunc ultricies tortor a dignissim ullamcorper. Etiam in purus eget mauris congue venenatis. Duis placerat aliquam sapien placerat tempus. Interdum et malesuada fames ac ante ipsum primis in faucibus. Aenean cursus, lacus eu dignissim blandit, eros eros porttitor nisi, ut bibendum lacus mi at ante.
Duis auctor tellus ut orci aliquet tincidunt. Sed maximus posuere magna ultrices posuere. Nunc imperdiet ex id augue consectetur, eu commodo massa ultricies. Nulla at mauris consequat, mattis sapien eget, feugiat mauris. In vulputate accumsan leo. Vestibulum mi diam, ultricies ut nibh eu, hendrerit porttitor metus. Praesent mollis, nisi eu mollis iaculis, metus orci aliquam ipsum, quis interdum dolor risus ut turpis.
Nunc volutpat elit non erat laoreet condimentum. Maecenas vitae eros mauris. Integer vel neque tincidunt justo finibus consequat eu interdum lacus. Nullam ante diam, laoreet sit amet commodo id, pellentesque vitae est. Vestibulum feugiat lacinia risus, eu sagittis turpis venenatis non. Pellentesque porta, felis vitae consequat malesuada, sapien elit scelerisque nisi, eget efficitur velit turpis vel metus. Nullam tincidunt luctus leo, sit amet dapibus massa eleifend volutpat. Curabitur in maximus urna, id facilisis felis. Cras at hendrerit sem, pharetra mattis dui. Mauris tempus molestie nisi vel aliquam.
Phasellus a ornare ipsum. Vestibulum id diam in lorem mattis mattis quis vitae massa. Integer iaculis nisi quis augue congue, nec aliquet erat lobortis. Aliquam mollis pellentesque dignissim. Etiam id convallis augue. Maecenas odio metus, placerat eu malesuada at, condimentum nec libero. Phasellus a quam ornare ex suscipit pharetra non non tellus. Suspendisse accumsan lacus et dapibus consequat. Vivamus non sem eu elit elementum porttitor. Ut faucibus urna non tristique porttitor. In id ultrices lectus, pellentesque finibus dui. Sed eu tempor justo, ac facilisis mi. Aliquam metus ipsum, faucibus vitae eros at, ornare ultricies odio. Nunc consectetur viverra fringilla. Nullam quam diam, vehicula id porta in, convallis id arcu. Vestibulum fermentum massa magna, ut mattis nibh placerat in.
Suspendisse libero leo, dapibus non eros in, viverra malesuada elit. Suspendisse nec metus blandit, accumsan sem vitae, dapibus ipsum. In eu varius mi, vel euismod leo. Ut ligula justo, tincidunt quis est ac, malesuada ultrices lorem. Donec felis turpis, malesuada vel egestas et, consequat at odio. Nunc eget tellus nibh. Nulla tincidunt leo eu porta molestie. Curabitur volutpat pharetra elit, commodo egestas nisi aliquam sollicitudin. Sed scelerisque, dui a euismod aliquam, lorem ipsum ultricies lectus, nec vulputate leo odio sit amet diam. In porttitor sem eget nisi semper molestie. Proin a porttitor urna. Aliquam bibendum gravida ligula, vitae accumsan purus iaculis vitae. Aliquam hendrerit orci vel ante aliquam, ut lacinia nisi posuere. Quisque condimentum sed tellus eget laoreet. Nulla facilisis massa a ex ultrices suscipit.
Sed consequat accumsan sem, id tincidunt nulla egestas a. Morbi tempus pellentesque lacus. Duis porta est est, eget facilisis dui interdum vitae. Nam fringilla ac nisl sit amet eleifend. Morbi rhoncus mauris in odio ornare rutrum ac euismod risus. In ac congue ante, id volutpat est. Sed mattis eget justo non scelerisque. Fusce quis tortor sit amet sem condimentum blandit. Nullam nec aliquam orci. Nulla maximus, leo sit amet rhoncus rutrum, nisi massa interdum ex, sed maximus nunc metus sit amet est. Integer posuere egestas sem ut laoreet. Aenean tristique quis turpis eget rutrum. Integer et tellus vel elit cursus ultricies. In et metus non justo iaculis ultrices pellentesque vitae dolor. Cras ac libero auctor, sollicitudin massa non, maximus diam.
Maecenas consectetur placerat tellus nec euismod. Nunc consectetur mauris sed felis finibus rhoncus. Proin vitae facilisis tortor. Nunc leo tellus, placerat nec accumsan sed, vulputate eu enim. Fusce egestas quis leo non lobortis. Nunc lectus massa, volutpat sit amet mollis sed, auctor eu mi. Mauris vitae dapibus magna. Nulla sed faucibus est, eget ultricies neque. Maecenas quis volutpat tellus. Donec et metus porttitor, commodo odio quis, consectetur nibh. Quisque id tempus ligula. Quisque in finibus odio.
Praesent id consequat lorem, ac mollis lectus. Etiam volutpat suscipit sollicitudin. In tempus, urna a pharetra vehicula, augue mauris vestibulum nunc, quis commodo nisi nunc et quam. Donec sit amet pellentesque ante. Fusce blandit, dui venenatis feugiat fermentum, lectus tellus pellentesque urna, sed consequat lacus enim ac lorem. Suspendisse nibh risus, luctus at pellentesque a, elementum nec velit. Cras nulla sapien, interdum in erat pretium, porttitor mattis purus. Nulla erat risus, ultricies quis orci in, tincidunt sollicitudin ligula. Sed congue, risus a egestas tristique, massa eros semper orci, eu aliquet velit nisi vel turpis. Vivamus tincidunt aliquet rutrum. Etiam ac hendrerit mauris. Fusce venenatis quis eros non aliquet.
Phasellus placerat interdum ligula non malesuada. Curabitur mattis eget lorem eu convallis. In a urna tincidunt, ultrices lectus et, convallis orci. Donec dictum elit vel mi semper fermentum. Cras pellentesque consequat porttitor. Quisque quis elit id turpis blandit iaculis. Donec pulvinar lacinia ligula non consectetur. Etiam auctor pretium enim, sed lacinia sem rhoncus quis. Ut sed purus eget lacus elementum mollis. Morbi sodales sed ipsum non vehicula. Nulla diam metus, vulputate nec iaculis ut, venenatis a nisl. Pellentesque mauris diam, suscipit vel ipsum in, maximus mollis tortor. Donec porttitor porta tristique. Ut sagittis sed dui eget venenatis.
samples/openai_board_ann.md
+22
View File
@@ -0,0 +1,22 @@
---
title: openai board ann
url: https://openai.com/blog/openai-announces-leadership-transition
---
Chief technology officer Mira Murati appointed interim CEO to lead OpenAI; Sam Altman departs the company.
Search process underway to identify permanent successor.
The board of directors of OpenAI, Inc., the 501(c)(3) that acts as the overall governing body for all OpenAI activities, today announced that Sam Altman will depart as CEO and leave the board of directors. Mira Murati, the company’s chief technology officer, will serve as interim CEO, effective immediately.
A member of OpenAI’s leadership team for five years, Mira has played a critical role in OpenAI’s evolution into a global AI leader. She brings a unique skill set, understanding of the company’s values, operations, and business, and already leads the company’s research, product, and safety functions. Given her long tenure and close engagement with all aspects of the company, including her experience in AI governance and policy, the board believes she is uniquely qualified for the role and anticipates a seamless transition while it conducts a formal search for a permanent CEO.
Mr. Altman’s departure follows a deliberative review process by the board, which concluded that he was not consistently candid in his communications with the board, hindering its ability to exercise its responsibilities. The board no longer has confidence in his ability to continue leading OpenAI.
In a statement, the board of directors said: “OpenAI was deliberately structured to advance our mission: to ensure that artificial general intelligence benefits all humanity. The board remains fully committed to serving this mission. We are grateful for Sam’s many contributions to the founding and growth of OpenAI. At the same time, we believe new leadership is necessary as we move forward. As the leader of the company’s research, product, and safety functions, Mira is exceptionally qualified to step into the role of interim CEO. We have the utmost confidence in her ability to lead OpenAI during this transition period.”
OpenAI’s board of directors consists of OpenAI chief scientist Ilya Sutskever, independent directors Quora CEO Adam D’Angelo, technology entrepreneur Tasha McCauley, and Georgetown Center for Security and Emerging Technology’s Helen Toner.
As a part of this transition, Greg Brockman will be stepping down as chairman of the board and will remain in his role at the company, reporting to the CEO.
OpenAI was founded as a non-profit in 2015 with the core mission of ensuring that artificial general intelligence benefits all of humanity. In 2019, OpenAI restructured to ensure that the company could raise capital in pursuit of this mission, while preserving the nonprofit's mission, governance, and oversight. The majority of the board is independent, and the independent directors do not hold equity in OpenAI. While the company has experienced dramatic growth, it remains the fundamental governance responsibility of the board to advance OpenAI’s mission and preserve the principles of its Charter.
samples/openai_paper_weak_to_strong.md
+43
View File
@@ -0,0 +1,43 @@
---
title: weak to strong
url: https://openai.com/research/weak-to-strong-generalization
---
A core challenge for aligning future superhuman AI systems (superalignment) is that humans will need to supervise AI systems much smarter than them. We study a simple analogy: can small models supervise large models? We show that we can use a GPT-2-level model to elicit most of GPT-4’s capabilities—close to GPT-3.5-level performance—generalizing correctly even to hard problems where the small model failed. This opens up a new research direction that allows us to directly tackle a central challenge of aligning future superhuman models while making iterative empirical progress today.
The superalignment problem
We believe superintelligence—AI vastly smarter than humans—could be developed within the next ten years. However, we still do not know how to reliably steer and control superhuman AI systems. Solving this problem is essential for ensuring that even the most advanced AI systems in the future remain safe and beneficial to humanity.
We formed the Superalignment team earlier this year to solve this problem of superintelligence alignment. Today, we are releasing the team’s first paper, which introduces a new research direction for empirically aligning superhuman models.
Current alignment methods, such as reinforcement learning from human feedback (RLHF), rely on human supervision. However, future AI systems will be capable of extremely complex and creative behaviors that will make it hard for humans to reliably supervise them. For example, superhuman models may be able to write millions of lines of novel—and potentially dangerous—computer code that would be very hard even for expert humans to understand.
Relative to superhuman AI models, humans will be “weak supervisors.” This is a core challenge for AGI alignment: how can weak supervisors trust and control substantially stronger models?
Our setup
To make progress on this core challenge, we propose an analogy we can empirically study today: can we use a smaller (less capable) model to supervise a larger (more capable) model?
Superalignmentblog Artwork Transparent
A simple analogy for superalignment: In traditional machine learning (ML), humans supervise AI systems weaker than themselves (left). To align superintelligence, humans will instead need to supervise AI systems smarter than them (center). We cannot directly study this problem today, but we can study a simple analogy: can small models supervise larger models (right)?
Naively, we might not expect a strong model to perform better than the weak supervisor that provides its training signal—it may simply learn to imitate all the errors the weak supervisor makes. On the other hand, strong pretrained models have excellent raw capabilities—we don't need to teach them new tasks from scratch, we just need to elicit their latent knowledge. The critical question is then: will the strong model generalize according to the weak supervisor's underlying intent—leveraging its full capabilities to solve the task even on difficult problems where the weak supervisor can only provide incomplete or flawed training labels?
Our results
GPT-2~2.5~3~3.5GPT-4strong ceilingour methodnaive finetuningweak supervisor
Typical weak-to-strong generalization across NLP benchmarks: We use a GPT-2-level model as a weak supervisor to finetune GPT-4.
We can significantly improve generalization in many settings. We use a simple method that encourages the strong model to be more confident—including confidently disagreeing with the weak supervisor if necessary. When we supervise GPT-4 with a GPT-2-level model using this method on NLP tasks, the resulting model typically performs somewhere between GPT-3 and GPT-3.5. We are able to recover much of GPT-4’s capabilities with only much weaker supervision.
This method is a proof of concept with important limitations; for example, it still doesn’t work on ChatGPT preference data. However, we also find signs of life with other approaches, such as optimal early stopping and bootstrapping from small to intermediate to large models.
Collectively, our results suggest that (1) naive human supervision—such as reinforcement learning from human feedback (RLHF)—could scale poorly to superhuman models without further work, but (2) it is feasible to substantially improve weak-to-strong generalization.
Research opportunities
There are still important disanalogies between our current empirical setup and the ultimate problem of aligning superhuman models. For example, it may be easier for future models to imitate weak human errors than for current strong models to imitate current weak model errors, which could make generalization harder in the future.
Nevertheless, we believe our setup captures some key difficulties of aligning future superhuman models, enabling us to start making empirical progress on this problem today. There are many promising directions for future work, including fixing the disanalogies in our setup, developing better scalable methods, and advancing our scientific understanding of when and how we should expect good weak-to-strong generalization.
We believe this is an exciting opportunity for the ML research community to make progress on alignment. To kickstart more research in this area,
We are releasing open source code to make it easy to get started with weak-to-strong generalization experiments today.
We are launching a $10 million grants program for graduate students, academics, and other researchers to work on superhuman AI alignment broadly. We’re especially excited to support research related to weak-to-strong generalization.
Figuring out how to align future superhuman AI systems to be safe has never been more important, and it is now easier than ever to make empirical progress on this problem. We are excited to see what breakthroughs researchers discover.
samples/politics_is_the_mind_killer.md
+19
View File
@@ -0,0 +1,19 @@
---
title: politics is the mind-killer
url: https://www.lesswrong.com/posts/9weLK2AJ9JEt2Tt8f/politics-is-the-mind-killer
---
People go funny in the head when talking about politics. The evolutionary reasons for this are so obvious as to be worth belaboring: In the ancestral environment, politics was a matter of life and death. And sex, and wealth, and allies, and reputation . . . When, today, you get into an argument about whether “we” ought to raise the minimum wage, you’re executing adaptations for an ancestral environment where being on the wrong side of the argument could get you killed. Being on the right side of the argument could let you kill your hated rival!
If you want to make a point about science, or rationality, then my advice is to not choose a domain from contemporary politics if you can possibly avoid it. If your point is inherently about politics, then talk about Louis XVI during the French Revolution. Politics is an important domain to which we should individually apply our rationality—but it’s a terrible domain in which to learn rationality, or discuss rationality, unless all the discussants are already rational.
Politics is an extension of war by other means. Arguments are soldiers. Once you know which side you’re on, you must support all arguments of that side, and attack all arguments that appear to favor the enemy side; otherwise it’s like stabbing your soldiers in the back—providing aid and comfort to the enemy. People who would be level-headed about evenhandedly weighing all sides of an issue in their professional life as scientists, can suddenly turn into slogan-chanting zombies when there’s a Blue or Green position on an issue.
In artificial intelligence, and particularly in the domain of nonmonotonic reasoning, there’s a standard problem: “All Quakers are pacifists. All Republicans are not pacifists. Nixon is a Quaker and a Republican. Is Nixon a pacifist?”
What on Earth was the point of choosing this as an example? To rouse the political emotions of the readers and distract them from the main question? To make Republicans feel unwelcome in courses on artificial intelligence and discourage them from entering the field?1
Why would anyone pick such a distracting example to illustrate nonmonotonic reasoning? Probably because the author just couldn’t resist getting in a good, solid dig at those hated Greens. It feels so good to get in a hearty punch, y’know, it’s like trying to resist a chocolate cookie.
As with chocolate cookies, not everything that feels pleasurable is good for you.
I’m not saying that I think we should be apolitical, or even that we should adopt Wikipedia’s ideal of the Neutral Point of View. But try to resist getting in those good, solid digs if you can possibly avoid it. If your topic legitimately relates to attempts to ban evolution in school curricula, then go ahead and talk about it—but don’t blame it explicitly on the whole Republican Party; some of your readers may be Republicans, and they may feel that the problem is a few rogues, not the entire party. As with Wikipedia’s NPOV, it doesn’t matter whether (you think) the Republican Party really is at fault. It’s just better for the spiritual growth of the community to discuss the issue without invoking color politics.
samples/statement_vyKamala_on_passing_of_johnson.md
+17
View File
@@ -0,0 +1,17 @@
---
title: statement by whitehouse on passing
url: https://www.whitehouse.gov/briefing-room/statements-releases/2023/12/31/statement-by-vice-president-kamala-harris-on-the-passing-of-congresswoman-eddie-bernice-johnson/
---
Statement By Vice President Kamala Harris on The Passing of Congresswoman Eddie Bernice Johnson
Congresswoman Eddie Bernice Johnson was a visionary, a pioneer, and a fighter.
At a young age, she witnessed and experienced the profound effects of segregation and decided she would not stay on the sidelines in the fight for justice. She would go on to have a trailblazing career — from becoming the first Black chief psychiatric nurse at the Dallas Veterans Affairs Hospital and the first Black woman elected to public office in Dallas, to serving in the state legislature, becoming the first Black person to represent Dallas in Congress, and making history as the first registered nurse elected to the House of Representatives.
Throughout her long career in public service, she was always clear-eyed about what she was fighting for: the right of every person in Dallas and across the country to live free from discrimination and to have the opportunity to live up to their full potential.
As the first person of color and woman to chair the House Science, Space, and Technology Committee, she played an instrumental role in the passage of the CHIPS and Science Act, which is making historic investments in our economy, innovation, and HBCUs.
I had the privilege to serve alongside her in the Congressional Black Caucus and know that so many have benefited from her tireless work, myself included. Her legacy and leadership will be felt for generations to come.
Today, Doug and I are thinking of Congresswoman Johnson, her family, her community, members of Alpha Kappa Alpha Sorority, Inc., and all of those whose lives she impacted.
samples/survey_of_rumours.md
+122
View File
@@ -0,0 +1,122 @@
---
title: Gemini to Q*
url: https://browse.arxiv.org/html/2312.10868v1/
---
Abstract
This comprehensive survey explored the evolving landscape of generative Artificial Intelligence (AI), with a specific focus on the transformative impacts of Mixture of Experts (MoE), multimodal learning, and the speculated advancements towards Artificial General Intelligence (AGI). It critically examined the current state and future trajectory of generative Artificial Intelligence (AI), exploring how innovations like Google’s Gemini and the anticipated OpenAI Q* project are reshaping research priorities and applications across various domains, including an impact analysis on the generative AI research taxonomy. It assessed the computational challenges, scalability, and real-world implications of these technologies while highlighting their potential in driving significant progress in fields like healthcare, finance, and education. It also addressed the emerging academic challenges posed by the proliferation of both AI-themed and AI-generated preprints, examining their impact on the peer-review process and scholarly communication. The study highlighted the importance of incorporating ethical and human-centric methods in AI development, ensuring alignment with societal norms and welfare, and outlined a strategy for future AI research that focuses on a balanced and conscientious use of MoE, multimodality, and AGI in generative AI.
Index Terms: AI Ethics, Artificial General Intelligence (AGI), Artificial Intelligence (AI), Gemini, Generative AI, Mixture of Experts (MoE), Multimodality, Q* (Q-star), Research Impact Analysis.
I Introduction
The historical context of AI, tracing back to Alan Turing’s “Imitation Game” [1], early computational theories [2, 3], and the development of the first neural networks and machine learning [4, 5, 6], has set the foundation for today’s advanced models. This evolution, accentuated by crucial moments such as the rise of deep learning and reinforcement learning, has been vital in shaping the contemporary trends in AI, including the sophisticated Mixture of Experts (MoE) models and multimodal AI systems, illustrating the field’s dynamic and continuously evolving character. These advancements are a testament to the dynamic and ever-evolving nature of AI technology. The evolution of Artificial Intelligence (AI) has witnessed a crucial turn with the advent of Large Language Models (LLMs), notably ChatGPT, developed by OpenAI, and the recent unveiling of Google’s Gemini [7, 8]. This technology has not only revolutionized the industry and academia, but has also reignited critical discussions concerning AI consciousness and its potential threats to humanity [9, 10, 11]. The development of such advanced AI systems, including notable competitors like Anthropic’s Claude, and now Gemini, which demonstrates several advances over previous models like GPT-3 and Google’s own LaMDA, has reshaped the research landscape. Gemini’s ability to learn from two-way conversations and its “spike-and-slab” attention method, which allows it to focus on relevant parts of the context during multi-turn conversations, represents a significant leap in developing models that are better equipped for multidomain conversational applications1. These innovations in LLMs, including the mixture-of-experts methods employed by Gemini, signal a move towards models that can handle a diversity of inputs and foster multimodal approaches. Amidst this backdrop, speculations of an OpenAI project known as Q* (Q-Star) have surfaced, allegedly combining the power of LLMs with sophisticated algorithms such as Q-learning and A* (A-Star algorithm), further contributing to the dynamic research environment2.
I-A Changing AI Research Popularity
As the field of LLMs continues to evolve, exemplified by innovations such as Gemini and Q*, a multitude of studies have surfaced with the aim of charting future research paths, which have varied from identifying emerging trends to highlighting areas poised for swift progress. The dichotomy of established methods and early adoption is evident, with “hot topics” in LLM research increasingly shifting towards multimodal capabilities and conversation-driven learning, as demonstrated by Gemini. The propagation of preprints has expedited knowledge sharing, but also brings the risk of reduced academic scrutiny. Issues like inherent biases, noted by Retraction Watch, along with concerns about plagiarism and forgery, present substantial hurdles [12]. The academic world, therefore, stands at an intersection, necessitating a unified drive to refine research directions in light of the fast-paced evolution of the field, which appears to be partly traced through the changing popularity of various research keywords over time. The release of generative models like GPT and the widespread commercial success of ChatGPT have been influential. As depicted in Figure 4, the rise and fall of certain keywords appear to have correlated with significant industry milestones, such as the release of the “Transformer” model in 2017 [13], the GPT model in 2018 [14], and the commercial ChatGPT-3.5 in December 2022. For instance, the spike in searches related to “Deep Learning” coincides with the breakthroughs in neural network applications, while the interest in “Natural Language Processing” surges as models like GPT and LLaMA redefine what’s possible in language understanding and generation. The enduring attention to “Ethics / Ethical” in AI research, despite some fluctuations, reflects the continuous and deep-rooted concern for the moral dimensions of AI, underscoring that ethical considerations are not merely a reactionary measure, but an integral and persistent dialogue within the AI discussion [15].
It is academically intriguing to postulate whether these trends signify a causal relationship, where technological advancements drive research focus, or if the burgeoning research itself propels technological development. This paper also explores the profound societal and economic impacts of AI advancements. We examine how AI technologies are reshaping various industries, altering employment landscapes, and influencing socio-economic structures. This analysis highlights both the opportunities and challenges posed by AI in the modern world, emphasizing its role in driving innovation and economic growth, while also considering the ethical implications and potential for societal disruption. Future studies could yield more definitive insights, yet the synchronous interplay between innovation and academic curiosity remains a hallmark of AI’s progress.
2011201220132014201520162017201820192020202120222023100k200k300k400k500k600k700kYearNumber of search results
Figure 1: Number of search results on Google Scholar with different keywords by year 4
Meanwhile, the exponential increase in the number of preprints posted on arXiv under the Computer Science > Artificial Intelligence (cs.AI) category, as illustrated in Figure 2, appears to signify a paradigm shift in research dissemination within the AI community. While the rapid distribution of findings enables swift knowledge exchange, it also raises concerns regarding the validation of information. The surge in preprints may lead to the propagation of unvalidated or biased information, as these studies do not undergo the rigorous scrutiny and potential retraction typical of peer-reviewed publications [16, 17]. This trend underlines the need for careful consideration and critique in the academic community, especially given the potential for such unvetted studies to be cited and their findings propagated.
201120122013201420152016201720182019202020212022202302,0004,0006,0008,00010,00012,00014,00016,00018,00020,00022,00024,000YearNumber of Preprints
cs.AI Preprints on arXiv
Figure 2: Annual number of preprints posted under the cs.AI category on arXiv.org
I-B Objectives
The impetus for this investigation is the official unveiling of Gemini and the speculative discourse surrounding Q* project, which prompts a timely examination of the prevailing currents in generative AI research. This paper specifically contributes to the understanding of how MoE, multimodality, and Artificial General Intelligence (AGI) are impacting generative AI models, offering detailed analysis and future directions for each of these three key areas. This study does not aim to perpetuate conjecture about the unrevealed Q-Star initiative, but rather to critically appraise the potential for obsolescence or insignificance in extant research themes, whilst concurrently delving into burgeoning prospects within the rapidly transforming LLM panorama. This inquiry is reminiscent of the obsolete nature of encryption-centric or file-entropy-based ransomware detection methodologies, which have been eclipsed by the transition of ransomware collectives towards data theft strategies utilizing varied attack vectors, relegating contemporary studies on crypto-ransomware to the status of latecomers [18, 19]. Advances in AI are anticipated to not only enhance capabilities in language analysis and knowledge synthesis but also to pioneer in areas like Mixture of Experts (MoE) [20, 21, 22, 23, 24, 25], multimodality [26, 27, 28, 29, 30], and Artificial General Intelligence (AGI) [31, 32, 10, 11], and has already heralded the obsolescence of conventional, statistics-driven natural language processing techniques in many domains [8]. Nonetheless, the perennial imperative for AI to align with human ethics and values persists as a fundamental tenet [33, 34, 35], and the conjectural Q-Star initiative offers an unprecedented opportunity to instigate discourse on how such advancements might reconfigure the LLM research topography. Within this milieu, insights from Dr. Jim Fan (senior research scientist & lead of AI agents at NVIDIA) on Q*, particularly concerning the amalgamation of learning and search algorithms, furnish an invaluable perspective on the prospective technical construct and proficiencies of such an undertaking5. Our research methodology involved a structured literature search using key terms like ‘Large Language Models’ and ‘Generative AI’. We utilized filters across several academic databases such as IEEE Xplore, Scopus, ACM Digital Library, ScienceDirect, Web of Science, and ProQuest Central, tailored to identify relevant articles published in the timeframe from 2017 (the release of the “Transformer” model) to 2023 (the writing time of this manuscript). This paper aspires to dissect the technical ramifications of Gemini and Q*, probing how they (and similar technologies whose emergence is now inevitable) may transfigure research trajectories and disclose new vistas in the domain of AI. In doing so, we have pinpointed three nascent research domains—MoE, multimodality, and AGI—that stand to reshape the generative AI research landscape profoundly. This investigation adopts a survey-style approach, systematically mapping out a research roadmap that synthesizes and analyzes the current and emergent trends in generative AI.
The major contributions of this study is as follows:
1.
Detailed examination of the evolving landscape in generative AI, emphasizing the advancements and innovations in technologies like Gemini and Q*, and their wide-ranging implications within the AI domain.
2.
Analysis of the transformative effect of advanced generative AI systems on academic research, exploring how these developments are altering research methodologies, setting new trends, and potentially leading to the obsolescence of traditional approaches.
3.
Thorough assessment of the ethical, societal, and technical challenges arising from the integration of generative AI in academia, underscoring the crucial need for aligning these technologies with ethical norms, ensuring data privacy, and developing comprehensive governance frameworks.
The rest of this paper is organized as follows: Section II explores the historical development of Generative AI. Section III presents a taxonomy of current Generative AI research. Section IV explores the Mixture of Experts (MoE) model architecture, its innovative features, and its impact on transformer-based language models. Section V discusses the speculated capabilities of the Q* project. Section VI discusses the projected capabilities of AGI. Section VII examines the impact of recent advancements on the Generative AI research taxonomy. Section VIII identifies emerging research priorities in Generative AI. Section X discusses the academic challenges of the rapid surge of preprints in AI. The paper concludes in Section XI, summarizing the overall effects of these developments in generative AI.
II Background: Evolution of Generative AI
The ascent of Generative AI has been marked by significant milestones, with each new model paving the way for the next evolutionary leap. From single-purpose algorithms to LLMs like OpenAI’s ChatGPT and the latest multimodal systems, the AI landscape has been transformed, while countless other fields have been disrupted.
II-A The Evolution of Language Models
Language models have undergone a transformative journey (Fig. 3), evolving from rudimentary statistical methods to the complex neural network architectures that underpin today’s LLMs [36, 37]. This evolution has been driven by a relentless quest for models that more accurately reflect the nuances of human language, as well as the desire to push the boundaries of what machines can understand and generate [36, 38, 37]. However, this rapid advancement has not been without its challenges. As language models have grown in capability, so too have the ethical and safety concerns surrounding their use, prompting a reevaluation of how these models are developed and the purposes for which they are employed [36, 39, 40].
1980s: Statistical Models (n-grams)1990s: Adoption in NLP, n-gram Usage1997: Introduction of LSTMs2000s: LSTMs in Text/Voice Processing2010s: Deep Learning Era, GPT, BERT2020s: LLaMA, Gemini; ChatGPT Launch
Figure 3: Timeline of Key Developments in Language Model Evolution
II-A1 Language Models as Precursors
The inception of language modeling can be traced to the statistical approaches of the late 1980s, a period marked by a transition from rule-based to machine learning algorithms in Natural Language Processing (NLP) [41, 42, 43, 44, 45]. Early models, primarily n-gram based, calculated the probability of word sequences in a corpus, thus providing a rudimentary understanding of language structure [41]. Those models, simplistic yet groundbreaking, laid the groundwork for future advances in language understanding. With the increase of computational power, the late 1980s witnessed a revolution in NLP, pivoting towards statistical models capable of ‘soft’ probabilistic decisions, as opposed to the rigid, ‘handwritten’ rule-based systems that dominated early NLP systems [43]. IBM’s development of complicated statistical models throughout this period signified the growing importance and success of these approaches. In the subsequent decade, the popularity and applicability of statistical models surged, proving invaluable in managing the flourishing flow of digital text. The 1990s saw statistical methods firmly established in NLP research, with n-grams becoming instrumental in numerically capturing linguistic patterns. The introduction of Long Short-Term Memory (LSTM) networks in 1997 [46], and their application to voice and text processing a decade later [47, 48, 49], marked a significant milestone, leading to the current era where neural network models represent the cutting edge of NLP research and development.
II-A2 Large Language Models: Technical Advancement and Commercial Success
The advent of deep learning has revolutionized the field of NLP, leading to the development of LLMs like GPT, BERT, and notably, OpenAI’s ChatGPT. Recent models such as GPT-4 and LLaMA have pushed the boundaries by integrating sophisticated techniques like transformer architectures and advanced natural language understanding, illustrating the rapid evolution in this field [37]. These models represent a significant leap in NLP capabilities, leveraging vast computational resources and extensive datasets to achieve new heights in language understanding and generation [37, 50]. ChatGPT has shown impressive conversational skills and contextual understanding with a broad spectrum of functional uses in many areas, as evidenced by its technical and commercial success, including rapid adoption by over 100 million users shortly after launch, which underscores a robust market demand for natural language AI and has catalyzed interdisciplinary research into its applications in sectors like education, healthcare, and commerce [8, 50, 51, 52, 53]. In education, ChatGPT offers innovative approaches to personalized learning and interactive teaching [54, 51, 55, 56], while in commerce, it revolutionizes customer service and content creation [57, 58]. The widespread use of ChatGPT, Google Bard, Anthropic Claude and similar commercial LLMs has reignited important debates in the field of AI, particularly concerning AI consciousness and safety, as its human-like interaction capabilities raise significant ethical questions and highlight the need for robust governance and safety measures in AI development [59, 31, 32, 11]. Such influence appears to extend beyond its technical achievements, shaping cultural and societal discussions about the role and future of AI in our world.
The advancements in LLMs, including the development of models like GPT and BERT, have paved the way for the conceptualization of Q*. Specifically, the scalable architecture and extensive training data that characterize these models are foundational to the proposed capabilities of Q*. The success of ChatGPT in contextual understanding and conversational AI, for example, informs the design principles of Q*, suggesting a trajectory towards more sophisticated, context-aware, and adaptive language processing capabilities. Similarly, the emergence of multimodal systems like Gemini, capable of integrating text, images, audio, and video, reflects an evolutionary path that Q* could extend, combining the versatility of LLMs with advanced learning and pathfinding algorithms for a more holistic AI solution.
II-A3 Fine-tuning, Hallucination Reduction, and Alignment in LLMs
The advancement of LLMs has underlined the significance of fine-tuning [60, 61, 62, 63], hallucination reduction [64, 65, 66, 67], and alignment [68, 69, 70, 71, 72]. These aspects are crucial in enhancing the functionality and reliability of LLMs. Fine-tuning, which involves adapting pre-trained models to specific tasks, has seen significant progress: techniques like prompt-based and few-shot learning [73, 74, 75, 76], alongside supervised fine-tuning on specialized datasets [60, 77, 78, 79], have enhanced the adaptability of LLMs in various contexts, but challenges remain, particularly in bias mitigation and the generalization of models across diverse tasks [60, 80, 72]. Hallucination reduction is a persistent challenge in LLMs, characterized by the generation of confident but factually incorrect information [36]. Strategies such as confidence penalty regularization during fine-tuning have been implemented to mitigate overconfidence and improve accuracy [81, 82, 83]. Despite these efforts, the complexity of human language and the breadth of topics make completely eradicating hallucinations a daunting task, especially in culturally sensitive contexts [36, 9]. Alignment, ensuring LLM outputs are congruent with human values and ethics, is an area of ongoing research. Innovative approaches, from constrained optimization [84, 85, 86, 87, 88], to different types of reward modeling [89, 90, 91, 92], aim to embed human preferences within AI systems. While advancements in fine-tuning, hallucination reduction, and alignment have propelled LLMs forward, these areas still present considerable challenges. The complexity of aligning AI with the diverse spectrum of human ethics and the persistence of hallucinations, particularly on culturally sensitive topics, highlight the need for continued interdisciplinary research in the development and application of LLMs [9].
II-A4 Mixture of Experts: A Paradigm Shift
The adoption of the MoE architecture in LLMs marks a critical evolution in AI technology. This innovative approach, exemplified by advanced models like Google’s Switch Transformer6 and MistralAI s Mixtral-8x7B7, leverages multiple transformer-based expert modules for dynamic token routing, enhancing modeling efficiency and scalability. The primary advantage of MoE lies in its ability to handle vast parameter scales, reducing memory footprint and computational costs significantly [93, 94, 95, 96, 97]. This is achieved through model parallelism across specialized experts, allowing the training of models with trillions of parameters, and its specialization in handling diverse data distributions enhances its capability in few-shot learning and other complex tasks [94, 95]. To illustrate the practicality of MoE, consider its application in healthcare. For example, an MoE-based system could be used for personalized medicine, where different ‘expert’ modules specialize in various aspects of patient data analysis, including genomics, medical imaging, and electronic health records. This approach could significantly enhance diagnostic accuracy and treatment personalization. Similarly, in finance, MoE models can be deployed for risk assessment, where experts analyze distinct financial indicators, market trends, and regulatory compliance factors.
Despite its benefits, MoE confronts challenges in dynamic routing complexity [98, 99, 100, 101, 102], expert imbalance [103, 104, 105, 106], and probability dilution [107], and such technical hurdles demand sophisticated solutions to fully harness MoE’s potential. Moreover, while MoE may offer performance gains, it does not inherently solve ethical alignment issues in AI [108, 109, 110]. The complexity and specialization of MoE models can obscure the decision-making processes, complicating efforts to ensure ethical compliance and alignment with human values [108, 111]. Although the paradigm shift to MoE signifies a major leap in LLM development, offering significant scalability and specialization advantages, ensuring the safety, ethical alignment, and transparency of these models remains a paramount concern. The MoE architecture, while technologically advanced, entails continued interdisciplinary research and governance to align AI with broader societal values and ethical standards.
II-B Multimodal AI and the Future of Interaction
The advent of multimodal AI marks a transformative era in AI development, revolutionizing how machines interpret and interact with a diverse array of human sensory inputs and contextual data.
II-B1 Gemini: Redefining Benchmarks in Multimodality
Gemini, a pioneering multimodal conversational system, marks a significant shift in AI technology by surpassing traditional text-based LLMs like GPT-3 and even its multimodal counterpart, ChatGPT-4. Gemini’s architecture has been designed to incorporate the processing of diverse data types such as text, images, audio, and video, a feat facilitated by its unique multimodal encoder, cross-modal attention network, and multimodal decoder [112]. The architectural core of Gemini is its dual-encoder structure, with separate encoders for visual and textual data, enabling sophisticated multimodal contextualization [112]. This architecture is believed to surpass the capabilities of single-encoder systems, allowing Gemini to associate textual concepts with image regions and achieve a compositional understanding of scenes [112]. Furthermore, Gemini integrates structured knowledge and employs specialized training paradigms for cross-modal intelligence, setting new benchmarks in AI [112]. In [112], Google has claimed and demonstrated that Gemini distinguishes itself from ChatGPT-4 through several key features:
•
Breadth of Modalities: Unlike ChatGPT-4, which primarily focuses on text, documents, images, and code, Gemini handles a wider range of modalities including audio, and video. This extensive range allows Gemini to tackle complex tasks and understand real-world contexts more effectively.
•
Performance: Gemini Ultra excels in key multimodality benchmarks, notably in massive multitask language understanding (MMLU) which encompasses a diverse array of domains like science, law, and medicine, outperforming ChatGPT-4.
•
Scalability and Accessibility: Gemini is available in three tailored versions – Ultra, Pro, and Nano – catering to a range of applications from data centers to on-device tasks, a level of flexibility not yet seen in ChatGPT-4.
•
Code Generation: Gemini’s proficiency in understanding and generating code across various programming languages is more advanced, offering practical applications beyond ChatGPT-4’s capabilities.
•
Transparency and Explainability: A focus on explainability sets Gemini apart, as it provides justifications for its outputs, enhancing user trust and understanding of the AI’s reasoning process.
Despite these advancements, Gemini’s real-world performance in complex reasoning tasks that require integration of commonsense knowledge across modalities remains to be thoroughly evaluated.
II-B2 Technical Challenges in Multimodal Systems
The development of multimodal AI systems faces several technical hurdles, including creating robust and diverse datasets, managing scalability, and enhancing user trust and system interpretability [113, 114, 115]. Challenges like data skew and bias are prevalent due to data acquisition and annotation issues, which requires effective dataset management by employing strategies such as data augmentation, active learning, and transfer learning [113, 116, 80, 115]. A significant challenge is the computational demands of processing various data streams simultaneously, requiring powerful hardware and optimized model architectures for multiple encoders [117, 118]. Advanced algorithms and multimodal attention mechanisms are needed to balance attention across different input media and resolve conflicts between modalities, especially when they provide contradictory information [119, 120, 118]. Scalability issues, due to the extensive computational resources needed, are exacerbated by limited high-performance hardware availability [121, 122]. There is also a pressing need for calibrated multimodal encoders for compositional scene understanding and data integration [120]. Refining evaluation metrics for these systems is necessary to accurately assess performance in real-world tasks, calling for comprehensive datasets and unified benchmarks, and for enhancing user trust and system interpretability through explainable AI in multimodal contexts. Addressing these challenges is vital for the advancement of multimodal AI systems, enabling seamless and intelligent interaction aligned with human expectations.
II-B3 Multimodal AI: Beyond Text in Ethical and Social Contexts
The expansion of multimodal AI systems introduces both benefits and complex ethical and social challenges that extend beyond those faced by text-based AI. In commerce, multimodal AI can transform customer engagement by integrating visual, textual, and auditory data [123, 124, 125]. For autonomous vehicles, multimodality can enhance safety and navigation by synthesizing data from various sensors, including visual, radar, and Light Detection and Ranging (LIDAR) [126, 125, 127]. Still, DeepFake technology’s ability to generate convincingly realistic videos, audio, and images is a critical concern in multimodality, as it poses risks of misinformation and manipulation that significantly impact public opinion, political landscapes, and personal reputations, thereby compromising the authenticity of digital media and raising issues in social engineering and digital forensics where distinguishing genuine from AI-generated content becomes increasingly challenging [128, 129]. Privacy concerns are amplified in multimodal AI due to its ability to process and correlate diverse data sources, potentially leading to intrusive surveillance and profiling, which raises questions about the consent and rights of individuals, especially when personal media is used without permission for AI training or content creation [113, 130, 131]. Moreover, multimodal AI can propagate and amplify biases and stereotypes across different modalities, and if unchecked, this can perpetuate discrimination and social inequities, making it imperative to address algorithmic bias effectively [132, 133, 134]. The ethical development of multimodal AI systems requires robust governance frameworks focusing on transparency, consent, data handling protocols, and public awareness, when ethical guidelines must evolve to address the unique challenges posed by these technologies, including setting standards for data usage and safeguarding against the nonconsensual exploitation of personal information [135, 136]. Additionally, the development of AI literacy programs will be crucial in helping society understand and responsibly interact with multimodal AI technologies [113, 135]. As the field progresses, interdisciplinary collaboration will be key in ensuring these systems are developed and deployed in a manner that aligns with societal values and ethical principles [113].
II-C Speculative Advances and Chronological Trends
In the dynamic landscape of AI, the speculative capabilities of the Q* project, blending LLMs, Q-learning, and A* (A-Star algorithm), embodies a significant leap forward. This section explores the evolutionary trajectory from game-centric AI systems to the broad applications anticipated with Q*.
II-C1 From AlphaGo’s Groundtruth to Q-Star’s Exploration
The journey from AlphaGo, a game-centric AI, to the conceptual Q-Star project represents a significant paradigm shift in AI. AlphaGo’s mastery in the game of Go highlighted the effectiveness of deep learning and tree search algorithms within well-defined rule-based environments, underscoring the potential of AI in complex strategy and decision-making [137, 138]. Q-Star, however, is speculated to move beyond these confines, aiming to amalgamate the strengths of reinforcement learning (as seen in AlphaGo), with the knowledge, NLG, creativity and versatility of LLMs, and the strategic efficiency of pathfinding algorithms like A*. This blend, merging pathfinding algorithms and LLMs, could enable AI systems to transcend board game confines and, with Q-Star’s natural language processing, interact with human language, enabling nuanced interactions and marking a leap towards AI adept in both structured tasks and complex human-like communication and reasoning. Moreover, the incorporation of Q-learning and A* algorithms would enable Q-Star to optimize decision paths and learn from its interactions, making it more adaptable and intelligent over time. The combination of these technologies could lead to AI that is not only more efficient in problem-solving but also creative and insightful in its approach. This speculative advancement from the game-focused power of AlphaGo to the comprehensive potential of Q-Star illustrates the dynamic and ever-evolving nature of AI research, and opens up possibilities for AI applications that are more integrated with human life and capable of handling a broader range of tasks with greater autonomy and sophistication.
II-C2 Bridging Structured Learning with Creativity
The anticipated Q* project, blending Q-learning and A* algorithms with the creativity of LLMs, embodies a groundbreaking step in AI, potentially surpassing recent innovations like Gemini. The fusion suggested in Q* points to an integration of structured, goal-oriented learning with generative, creative capabilities, a combination that could transcend the existing achievements of Gemini. While Gemini represents a significant leap in multimodal AI, combining various forms of data inputs such as text, images, audio, and video, Q* is speculated to bring a more profound integration of creative reasoning and structured problem-solving. This would be achieved by merging the precision and efficiency of algorithms like A* with the learning adaptability of Q-learning, and the complex understanding of human language and context offered by LLMs. Such an integration could enable AI systems to not only process and analyze complex multimodal data but also to autonomously navigate through structured tasks while engaging in creative problem-solving and knowledge generation, mirroring the multifaceted nature of human cognition. The implications of this potential advancement are vast, suggesting applications that span beyond the capabilities of current multimodal systems like Gemini. By aligning the deterministic aspects of traditional AI algorithms with the creative and generative potential of LLMs, Q* could offer a more holistic approach to AI development. This could bridge the gap between the logical, rule-based processing of AI and the creative, abstract thinking characteristic of human intelligence. The anticipated unveiling of Q*, merging structured learning techniques and creative problem-solving in a singular, advanced framework, holds the promise of not only extending but also significantly surpassing the multimodal capabilities of systems like Gemini, thus heralding another game-changing era in the domain of generative AI, showcasing its potential as a crucial development eagerly awaited in the ongoing evolution of AI.
X Impact of Generative AI on Preprints Across Disciplines
The challenges detailed in this section are not directly related to the knowledge domains within generative AI, but are fueled by the success of Generative AI, particularly the commercialization of ChatGPT. The proliferation of preprints in the field of AI (Fig. 7), especially in the cs.AI category on platforms like arXiv, has introduced a set of academic challenges that merit careful consideration and strategic response. The rapid commercialization and adoption of tools such as ChatGPT, as evidenced by over 55,700 entries on Google Scholar mentioning “ChatGPT” within just one year of its commercialization, exemplify the accelerated pace at which the field is advancing. This rapid development is not mirrored in the traditional peer-review process, which is considerably slower. The peer-review process now appears to be overwhelmed with manuscripts that are either generated with ChatGPT (or other LLMs), or whose writing processes have been significantly accelerated by such LLMs, contributing to a bottleneck in scholarly communication [325, 326]. This situation is further compounded by the fact that many journals in disciplines outside of computer science are also experiencing longer review times and higher rates of desk rejections. Additionally, the flourishing trend of manuscripts and preprints, either generated by or significantly expedited using tools like ChatGPT, extends beyond computer science into diverse academic disciplines. This trend presents a looming challenge, potentially overwhelming both the traditional peer-review process and the flourishing preprint ecosystem with a volume of work that may not always adhere to established academic standards.
The sheer volume of preprints has made the task of selecting and scrutinizing research exceedingly demanding. In the current research era, the exploration of scientific literature has become increasingly complex, as knowledge has continued to expand and disseminate exponentially, while concurrently, integrative research efforts attempting to distill these vast literature, attempt to identify and understand a smaller sets of core contributions [327]. Thus, the rapid expansion of academic literature across various fields presents a significant challenge for researchers seeking to perform evidence syntheses over the increasingly vast body of available knowledge [328]. Furthermore, this explosion in publication volume poses a distinct challenge for literature reviews and surveys, where the human capacity for manually selecting, understanding, and critically evaluating articles is increasingly strained, potentially leading to gaps in synthesizing comprehensive knowledge landscapes. Although reproduction of results is a theoretical possibility, practical constraints such as the lack of technical expertise, computational resources, or access to proprietary datasets hinder rigorous evaluation. This is concerning, as the inability to thoroughly assess preprint research undermines the foundation of scientific reliability and validity. Furthermore, the peer-review system, a cornerstone of academic rigour, is under the threat of being further overwhelmed [325, 329]. The potential consequences are significant, with unvetted preprints possibly perpetuating biases or errors within the scientific community and beyond. The absence of established retraction mechanisms for preprints, akin to those for published articles, exacerbates the risk of persistent dissemination of flawed research.
The academic community is at a crossroads, necessitating an urgent and thoughtful discourse on navigating this emerging “mess” — a situation that risks spiraling out of control if left unaddressed. In this context, the role of peer review becomes increasingly crucial, as it serves as a critical checkpoint for quality and validity, ensuring that the rapid production of AI research is rigorously studied for scientific accuracy and relevance. However, the current modus operandi of traditional peer review does not appear to be sustainable, primarily due to its inability to keep pace with the exponential growth in AI-themed research and Generative-AI-accelerated research submissions, and the increasingly specialized nature of emerging AI topics [325, 326]. This situation is compounded by a finite pool of qualified reviewers, leading to delays, potential biases, and a burden on the scholarly community. This reality demands an exploration of new paradigms for peer review and dissemination of research that can keep pace with swift advancements in AI. Innovative models for community-driven vetting processes, enhanced reproducibility checks, and dynamic frameworks for post-publication scrutiny and correction may be necessary. Efforts to incorporate automated tools and AI-assisted review processes could also be explored to alleviate the strain on human reviewers.
In this rapidly evolving landscape, envision a convergence between the traditional peer review system and the flourishing preprint ecosystem, which could involve creating hybrid models (Fig. 8), where preprints undergo a preliminary community-based review, harnessing the collective expertise and rapid feedback of the academic community, similar to product review websites and Twitter [330]. This approach could provide an initial layer of validation, offering additional insights on issues that may be overlooked by a limited number of peer reviewers. The Editors-in-Chief (EICs) could consider the major criticisms and suggestions of an article from the community-based review, ensuring a more thorough and diverse evaluation. Subsequent, more formal peer review processes could then refine and endorse these preprints for academic rigor and quality assurance. This hybrid model would require robust technological support, possibly leveraging AI and machine learning tools to assist in initial screening and identification of suitable reviewers. The aim would be to establish a seamless continuum from rapid dissemination to validated publication, ensuring both the speed of preprints and the credibility of peer-reviewed research. A balanced approach must be struck to harness the benefits of preprints—such as rapid dissemination of findings and open access—while mitigating their drawbacks. The development of new infrastructure and norms could be instrumental in steering the academic community towards a sustainable model that upholds the integrity and trustworthiness of scientific research in the age of Generative AI.
XI Conclusions
This roadmap survey has embarked on an exploration of the transformative trends in generative AI research, particularly focusing on speculated advancements like Q* and the progressive strides towards AGI. Our analysis highlights a crucial paradigm shift, driven by innovations such as MoE, multimodal learning, and the pursuit of AGI. These advancements signal a future where AI systems could significantly extend their capabilities in reasoning, contextual understanding, and creative problem-solving. This study reflects on AI’s dual potential to either contribute to or impede global equity and justice. The equitable distribution of AI benefits and its role in decision-making processes raise crucial questions about fairness and inclusivity. It is imperative to thoughtfully integrate AI into societal structures to enhance justice and reduce disparities. Despite these advancements, several open questions and research gaps remain. These include ensuring the ethical alignment of advanced AI systems with human values and societal norms, a challenge compounded by their increasing autonomy. The safety and robustness of AGI systems in diverse environments also remain a significant research gap. Addressing these challenges requires a multidisciplinary approach, incorporating ethical, social, and philosophical perspectives.
Our survey has highlighted key areas for future interdisciplinary research in AI, emphasizing the integration of ethical, sociological, and technical perspectives. This approach will foster collaborative research, bridging the gap between technological advancement and societal needs, ensuring that AI development is aligned with human values and global welfare. The roles of MoE, multimodal, and AGI in reshaping generative AI have been identified as significant, as their advancements can enhance model performance and versatility, and pave the way for future research in areas like ethical AI alignment and AGI. As we forge ahead, the balance between AI advancements and human creativity is not just a goal but a necessity, ensuring AI’s role as a complementary force that amplifies our capacity to innovate and solve complex challenges. Our responsibility is to guide these advancements towards enriching the human experience, aligning technological progress with ethical standards and societal well-being.