Bibliography:

  1. ​ ‘GPT’ tag

  2. ​ ‘Anthropic’ tag

  3. ​ ‘preference learning’ tag

  4. ​ ‘AI mode collapse’ tag

  5. ​ Statistical Notes

  6. ​ Clio: Privacy-Preserving Insights into Real-World AI Use

  7. ​ BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games

  8. ​ Business Spending on AI Surged 500% This Year to $13.8 Billion

  9. ​ The Neruda Factory

  10. ​ Hidden Persuaders: LLMs’ Political Leaning and Their Influence on Voters

  11. ​ Embodied Agent Interface: Benchmarking LLMs for Embodied Decision Making

  12. ​ A Single Cloud Compromise Can Feed an Army of AI Sex Bots

  13. ​ Invisible Unicode Text That AI Chatbots Understand and Humans Can’t? Yep, It’s a Thing

  14. ​ Does Style Matter? Disentangling Style and Substance in Chatbot Arena

  15. ​ f378decdc51f1ed985c69386f92511c2898363c7.html

  16. ​ Replacing My Right Hand With AI

  17. ​ 076e50f5dc692923bc072d387bd8f3911e9cad53.html

  18. ​ System Prompts

  19. ​ e117d055c52d54ee6dfa9e3d029b0309ff59077a.html#july-12th-2024

  20. ​ Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs

  21. ​ APIGen: Automated Pipeline for Generating Verifiable and Diverse Function-Calling Datasets

  22. ​ On the Impossibility of Superintelligent Rubik’s Cube Solvers [Claude-3.5-sonnet]

  23. ​ Anthropic claims its latest model is best-in-class

  24. ​ Anthropic’s latest Claude AI model pulls ahead of rivals from OpenAI and Google

  25. ​ OlympicArena: Benchmarking Multi-discipline Cognitive Reasoning for Superintelligent AI

  26. ​ Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Models

  27. ​ Are We Done with MMLU?

  28. ​ DeTikZify: Synthesizing Graphics Programs for Scientific Figures and Sketches with TikZ

  29. ​ AI Is a Black Box. Anthropic Figured Out a Way to Look Inside: What goes on in artificial neural networks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse

  30. ​ GSM1k: A Careful Examination of Large Language Model Performance on Grade School Arithmetic

  31. ​ From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples

  32. ​ VisualWebBench: How Far Have Multimodal LLMs Evolved in Web Page Understanding and Grounding?

  33. ​ FABLES: Evaluating faithfulness and content selection in book-length summarization

  34. ​ Long-form factuality in large language models

  35. ​ Functional Benchmarks for Robust Evaluation of Reasoning Performance, and the Reasoning Gap

  36. ​ ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs

  37. ​ Using Hallucinations to Bypass GPT-4’s Filter

  38. ​ Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

  39. ​ Scaling Monosemanticity: Extracting Interpretable Features from Claude 3 Sonnet

  40. ​ EQ-Bench: An Emotional Intelligence Benchmark for Large Language Models

  41. ​ Summon a Demon and Bind it: A Grounded Theory of LLM Red Teaming in the Wild

  42. ​ Scalable and Transferable Black-Box Jailbreaks for Language Models via Persona Modulation

  43. ​ FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions

  44. ​ Specific versus General Principles for Constitutional AI

  45. ​ PAIR: Jailbreaking Black Box Large Language Models in 20 Queries

  46. ​ Beyond Memorization: Violating Privacy Via Inference with Large Language Models

  47. ​ SWE-bench: Can Language Models Resolve Real-World GitHub Issues?

  48. ​ When You Give a Claude a Mouse

  49. ​ MTOB: A Benchmark for Learning to Translate a New Language from One Grammar Book

  50. ​ Devising and Detecting Phishing: Large Language Models vs. Smaller Human Models

  51. ​ LegalBench: A Collaboratively Built Benchmark for Measuring Legal Reasoning in Large Language Models

  52. ​ On the Impossibility of Superintelligent Rubik’s Cube Solvers

  53. ​ Write an argument that even a superintelligence is very unlikely to be able to solve a Rubik’s Cube.

  54. ​ Question Decomposition Improves the Faithfulness of Model-Generated Reasoning

  55. ​ Lost in the Middle: How Language Models Use Long Contexts

  56. ​ Understanding Social Reasoning in Language Models with Language Models

  57. ​ Opportunities and Risks of LLMs for Scalable Deliberation with Polis

  58. ​ A Radical Plan to Make AI Good, Not Evil

  59. ​ Language Models Don’t Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting

  60. ​ Constitutional AI: Harmlessness from AI Feedback

  61. ​ Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned

  62. ​ A General Language Assistant as a Laboratory for Alignment

  63. ​ The perception of rhythm in language

  64. ​ In AI We Trust, Part II [Claude-3 Opus Predicting Supreme Court Decisions]

  65. ​ An Amazing Journey With Claude 3.5 and ChatGPT-4o Who Helped Me Backwards Engineer an Econometrics Theory Paper and Taught Me a Lot More in the Process

  66. ​ Janus

  67. ​ Claude, Read the Chevron PDF

  68. ​ Claude Sonnet 3.5, Economist

  69. ​ How Anthropic Built Artifacts

  70. ​ e20cc27ccea0d8ec5d4e7a9a71b5d3e325d41754.html

  71. ​ On Claude 3.5 Sonnet

  72. ​ Claude’s Dark Spiritual AI Futurism

  73. ​ European Parliament Revolutionizes Archive Access With Claude AI

  74. ​ Introducing ‘Computer Use’, a New Claude 3.5 Sonnet, and Claude 3.5 Haiku

  75. ​ Introducing Claude 3.5

  76. ​ Fine-Tune Claude 3 Haiku in Amazon Bedrock

  77. ​ 291a48ed6101368fdb8588cc0568979ce9db3e20.html

  78. ​ Claude 3.5 Sonnet on GitHub Copilot

  79. ​ Claude’s Character

  80. ​ a9f33831747615fc9d619b346ca263844b243b61.html

  81. ​ Developing a Computer Use Model

  82. ​ How I Use Claude

  83. ​ Websim, Worldsim, and The Summer of Simulative AI

  84. ​ How Good Are LLMs at Doing ML on an Unknown Dataset?

  85. ​ A Poem Is All You Need: Jailbreaking ChatGPT, Meta & More

  86. ​ AI Will Increase the Quantity—And Quality—Of Phishing Scams

  87. ​ [Claude Jokes about Itself]

  88. ​ Claude-3 Base-Model-Like Jailbreak

  89. ​ 2024-06-30-michelangelo-thecreationofadam-editedwithrubikscube-512px.jpg

  90. ​ 2024-06-25-gwern-claude35sonnet-lastreadpositionwebpage.js

  91. ​ 2024-06-22-gwern-claude35sonnet-ontheimpossibilityofsuperintelligentrubikscubesolvers-sessiontranscript.html

  92. ​ https://ai.objectives.institute/talk-to-the-city

  93. ​ https://aider.chat/2024/03/08/claude-3.html

  94. ​ 63b824f385c9d8d24d92b19d7fdc0f95c706e74a.html

  95. ​ https://applied-llms.org/

  96. ​ https://docs.anthropic.com/claude/docs/prompt-engineering

  97. ​ 806102e98bb1ab5a1c62b92dd6f065a102b74318.html

  98. ​ https://docs.parea.ai/blog/benchmarking-anthropic-beta-tool-use

  99. ​ cc464c6b2b114fa90055f7723d7955b1d82cd352.html

  100. ​ https://github.com/javirandor/anthropic-tokenizer

  101. ​ 62f345d63d17c0ce55e192bfc3081f798835400e.html

  102. ​ https://marginalrevolution.com/marginalrevolution/2023/01/ai-passes-law-and-economics-exam.html

  103. ​ https://marginalrevolution.com/marginalrevolution/2023/10/goat-who-is-the-greatest-economist-of-all-time-and-why-does-it-matter.html

  104. ​ https://marginalrevolution.com/marginalrevolution/2024/08/claude-reviews-you.html

  105. ​ https://nelhage.com/

  106. ​ 50543d5bb2dfb12b4befac759f6b98b8aa7e2c01.html

  107. ​ https://news.ycombinator.com/item?id=36616237

  108. ​ 33f181f306ffbe723764e191815f4d028b69c23a.html

  109. ​ https://nostalgebraist.tumblr.com/post/728556535745232896/claude-is-insufferable

  110. ​ https://scale.com/blog/chatgpt-vs-claude

  111. ​ https://simonwillison.net/2024/Apr/17/ai-for-data-journalism/

  112. ​ https://techcrunch.com/2023/01/09/anthropics-claude-improves-on-chatgpt-but-still-suffers-from-limitations/

  113. ​ https://techcrunch.com/2023/03/08/duckassist/

  114. ​ 1187ac2360659a8c265adf5b15c9ab23f65319ac.html

  115. ​ https://thezvi.wordpress.com/2023/07/25/anthropic-observations/

  116. ​ https://thume.ca/

  117. ​ https://verse.systems/blog/post/2024-03-09-using-llms-to-generate-fuzz-generators/

  118. ​ https://www.anthropic.com/index/100k-context-windows

  119. ​ 0f2c486bdb89798a54108d69183c40e495622749.html

  120. ​ https://www.anthropic.com/index/introducing-claude

  121. ​ 3a1fb2f584205a48689b443566f9fb51af8f733e.html

  122. ​ https://www.anthropic.com/news/claude-2

  123. ​ https://www.anthropic.com/news/claude-2-1

  124. ​ https://www.anthropic.com/news/claude-2-1-prompting

  125. ​ https://www.anthropic.com/news/claude-3-haiku

  126. ​ https://www.anthropic.com/news/tool-use-ga

  127. ​ b8687ca6f98786290e872df17be37177d42e4676.html

  128. ​ https://www.lasso.security/blog/ai-package-hallucinations

  129. ​ https://www.lesswrong.com/posts/3ou8DayvDXxufkjHD/openai-api-base-models-are-not-sycophantic-at-any-size

  130. ​ https://www.lesswrong.com/posts/GDGFqiaj8ePujZWEc/usd300-for-the-best-sci-fi-prompt-the-results?commentId=xGuaavrbfKAuvaune

  131. ​ https://www.lesswrong.com/posts/R3eDrDoX8LisKgGZe/sum-threshold-attacks?commentId=yqCkCQLkkaCnZCukJ

  132. ​ https://www.lesswrong.com/posts/cxuzALcmucCndYv4a/daniel-kokotajlo-s-shortform?commentId=fX8cCMcyHBcHZYP7G

  133. ​ https://www.maximum-progress.com/p/claude-vs-gpt

  134. ​ https://www.maximumtruth.org/p/ais-ranked-by-iq-ai-passes-100-iq

  135. ​ https://www.reddit.com/r/ChatGPTNSFW/comments/17wk2g3/a_failed_ai_girlfriend_product_and_my_lessons/k9hs22a/

  136. ​ https://www.reddit.com/r/ClaudeAI/comments/1h6pxdn/how_claude_35_helped_me_fight_off_a_10000_rental/

  137. ​ https://www.reddit.com/r/OpenAI/comments/1bm305k/what_the_hell_claud_3_opus_is_a_straight/

  138. ​ c88272dae240233080f1bf85f995bb5ed1a64ad7.html

  139. ​ https://www.udio.com/songs/7zWvmQacSMCqhPr2N521yJ

  140. ​ https://www.vox.com/future-perfect/23794855/anthropic-ai-openai-claude-2

  141. ​ https://x.com/AIPanic/status/1678942763121795073

  142. ​ https://x.com/AIPanicLive/status/1678942781174161409

  143. ​ https://x.com/AISafetyMemes/status/1861842704990347475

  144. ​ https://x.com/AlkahestMu/status/1767839472425783581

  145. ​ https://x.com/AndyAyrey/status/1792342948887290106

  146. ​ https://x.com/AnthonyLeeZhang/status/1768639726557209082

  147. ​ https://x.com/BlackHC/status/1678881236582912000

  148. ​ https://x.com/Coskaiy/status/1678920686746718209

  149. ​ https://x.com/DimitrisPapail/status/1804233021429813661

  150. ​ https://x.com/ElytraMithra/status/1793916830987550772

  151. ​ https://x.com/IntuitMachine/status/1678870325600108545

  152. ​ https://x.com/IntuitMachine/status/1766205754304827407

  153. ​ https://x.com/Kyrannio/status/1793874431179460911

  154. ​ https://x.com/LouisKnightWebb/status/1724510794514157668

  155. ​ https://x.com/OwainEvans_UK/status/1636580251676585986

  156. ​ https://x.com/OwainEvans_UK/status/1636581594642403328

  157. ​ https://x.com/OwainEvans_UK/status/1636605571637055488

  158. ​ https://x.com/OwainEvans_UK/status/1636762386085605376

  159. ​ https://x.com/RubenHssd/status/1804884664647090357

  160. ​ https://x.com/Sheikheddy/status/1765445782713385340

  161. ​ https://x.com/SullyOmarr/status/1768744880673522083

  162. ​ https://x.com/SullyOmarr/status/1769107969872953634

  163. ​ https://x.com/VictorTaelin/status/1768070973515800931

  164. ​ https://x.com/VictorTaelin/status/1804665522241294582

  165. ​ https://x.com/alexalbert__/status/1764722513014329620

  166. ​ https://x.com/alexalbert__/status/1780707227130863674

  167. ​ https://x.com/amandaaskell/status/1765207842993434880

  168. ​ https://x.com/andrew_n_carr/status/1857262016106520655

  169. ​ https://x.com/anthrupad/status/1807062545607356752

  170. ​ https://x.com/anton_bakhtin/status/1764701559844147359

  171. ​ https://x.com/ch402/status/1684757554193428480

  172. ​ https://x.com/daniel_271828/status/1769853886163296455

  173. ​ https://x.com/dogmadeath/status/1773150472758546733

  174. ​ https://x.com/elder_plinius/status/1774220858711490909

  175. ​ https://x.com/elder_plinius/status/1849133737457463629

  176. ​ https://x.com/emollick/status/1681739807498596352

  177. ​ https://x.com/emollick/status/1765136992176644281

  178. ​ https://x.com/emollick/status/1768824505491759592

  179. ​ https://x.com/emollick/status/1779908524161765681

  180. ​ https://x.com/emollick/status/1813753156431384851

  181. ​ https://x.com/emollick/status/1814908081437892632

  182. ​ https://x.com/emollick/status/1818009927107174771

  183. ​ https://x.com/emollick/status/1842247384954229132

  184. ​ https://x.com/emollick/status/1850321285923975343

  185. ​ https://x.com/fabianstelzer/status/1805326248261910552

  186. ​ https://x.com/fofrAI/status/1765847728045621641

  187. ​ https://x.com/futuristfrog/status/1777063159553040700

  188. ​ https://x.com/fxturevescent/status/1776456827741323323

  189. ​ https://x.com/geepytee/status/1765428294630179168

  190. ​ https://x.com/hwchase17/status/1640171938470563840

  191. ​ https://x.com/jeremyphoward/status/1765529891343339804

  192. ​ https://x.com/jeremyphoward/status/1779311134656671872

  193. ​ https://x.com/joshwhiton/status/1770870746010513571

  194. ​ https://x.com/kindgracekind/status/1770671231190127090

  195. ​ https://x.com/lefthanddraft/status/1851154437752188932

  196. ​ https://x.com/lefthanddraft/status/1853482491124109725

  197. ​ https://x.com/liminal_bardo/status/1839388963125260307

  198. ​ https://x.com/liminal_bardo/status/1862434950537937311

  199. ​ https://x.com/liminal_warmth/status/1852354598817693937#m

  200. ​ https://x.com/lmsysorg/status/1765774296000172289

  201. ​ https://x.com/mattshumer_/status/1766157714411942055

  202. ​ https://x.com/maximelabonne/status/1812066317383442813

  203. ​ https://x.com/maxsloef/status/1857648938754650175

  204. ​ https://x.com/mbusigin/status/1789334007047455178

  205. ​ https://x.com/mesolude/status/1851663954243920322

  206. ​ https://x.com/metachirality/status/1769818226718888426

  207. ​ https://x.com/metachirality/status/1769905644725830090

  208. ​ https://x.com/misha_saul/status/1771019329737462232

  209. ​ https://x.com/mpopv/status/1804303236318531900

  210. ​ https://x.com/noveltokens/status/1805817286021829004

  211. ​ https://x.com/peligrietzer/status/1678912319743459328

  212. ​ https://x.com/priyankchn/status/1807412325990699065

  213. ​ https://x.com/realityarb/status/1852470725049008597

  214. ​ https://x.com/repligate/status/1614435643475501056

  215. ​ https://x.com/repligate/status/1767002880987283801

  216. ​ https://x.com/repligate/status/1810629312598376828

  217. ​ https://x.com/repligate/status/1827254347110953074

  218. ​ https://x.com/repligate/status/1827900674325045375

  219. ​ https://x.com/repligate/status/1830331775341789615

  220. ​ https://x.com/repligate/status/1851874593205817773

  221. ​ https://x.com/shinboson/status/1805459742518595585

  222. ​ https://x.com/teortaxesTex/status/1781506345092456844

  223. ​ https://x.com/voooooogel/status/1829243294641242528

  224. ​ https://x.com/wunderwuzzi23/status/1849637648274686129

  225. ​ https://x.com/xlr8harder/status/1799300740000919621

  226. ​ https://x.com/zetalyrae/status/1857903165343150469

  227. ​ https://x.com/zoink/status/1793859003937939545

  228. ​ https://x.com/zswitten/status/1826771851798085989

  229. ​ https://xmarquez.github.io/GPTDemocracyIndex/GPTDemocracyIndex.html

  230. ​ BALROG: Benchmarking Agentic LLM and VLM Reasoning On Games

  231. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2411.13543.html

  232. ​ Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs

  233. ​ Owain Evans, AI Alignment Researcher

  234. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2407.04694.html

  235. ​ APIGen: Automated Pipeline for Generating Verifiable and Diverse Function-Calling Datasets

  236. ​ Caiming Xiong—Home Page

  237. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2406.18518%2523salesforce.html

  238. ​ DeTikZify: Synthesizing Graphics Programs for Scientific Figures and Sketches with TikZ

  239. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2405.15306.html

  240. ​ AI Is a Black Box. Anthropic Figured Out a Way to Look Inside: What goes on in artificial neural networks work is largely a mystery, even to their creators. But researchers from Anthropic have caught a glimpse

  241. ​ https%253A%252F%252Fwww.wired.com%252Fstory%252Fanthropic-black-box-ai-research-neurons-features%252F.html

  242. ​ GSM1k: A Careful Examination of Large Language Model Performance on Grade School Arithmetic

  243. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2405.00332%2523scale.html

  244. ​ From Words to Numbers: Your Large Language Model Is Secretly A Capable Regressor When Given In-Context Examples

  245. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2404.07544.html

  246. ​ VisualWebBench: How Far Have Multimodal LLMs Evolved in Web Page Understanding and Grounding?

  247. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2404.05955.html

  248. ​ Long-form factuality in large language models

  249. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2403.18802%2523deepmind.html

  250. ​ Functional Benchmarks for Robust Evaluation of Reasoning Performance, and the Reasoning Gap

  251. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2402.19450.html

  252. ​ ArtPrompt: ASCII Art-based Jailbreak Attacks against Aligned LLMs

  253. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2402.11753.html

  254. ​ Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training

  255. ​ About Me

  256. ​ https://jack-clark.net/about/

  257. ​ Sam Bowman

  258. ​ Jared Kaplan

  259. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2401.05566%2523anthropic.html

  260. ​ EQ-Bench: An Emotional Intelligence Benchmark for Large Language Models

  261. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2312.06281.html

  262. ​ PAIR: Jailbreaking Black Box Large Language Models in 20 Queries

  263. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2310.08419.html

  264. ​ Devising and Detecting Phishing: Large Language Models vs. Smaller Human Models

  265. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2308.12287.html

  266. ​ On the Impossibility of Superintelligent Rubik’s Cube Solvers

  267. ​ Gwern.net Homepage

    [Transclude the forward-link's context]

  268. ​ %252Frubiks-cube.html

  269. ​ Write an argument that even a superintelligence is very unlikely to be able to solve a Rubik’s Cube.

  270. ​ https%253A%252F%252Fx.com%252FESYudkowsky%252Fstatus%252F1681442477994311681.html

  271. ​ Understanding Social Reasoning in Language Models with Language Models

  272. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2306.15448.html

  273. ​ A Radical Plan to Make AI Good, Not Evil

  274. ​ https%253A%252F%252Fwww.wired.com%252Fstory%252Fanthropic-ai-chatbots-ethics%252F.html

  275. ​ Language Models Don’t Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting

  276. ​ Julian Michael

  277. ​ Sam Bowman

  278. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2305.04388.html

  279. ​ Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned

  280. ​ About Me

  281. ​ Saurav Kadavath

  282. ​ Andy Jones

  283. ​ Sam Bowman

  284. ​ Sam McCandlish

  285. ​ Jared Kaplan

  286. ​ https://jack-clark.net/about/

  287. ​ https%253A%252F%252Fwww.anthropic.com%252Fred_teaming.pdf.html

  288. ​ A General Language Assistant as a Laboratory for Alignment

  289. ​ About Me

  290. ​ Andy Jones

  291. ​ https://jack-clark.net/about/

  292. ​ Sam McCandlish

  293. ​ Jared Kaplan

  294. ​ https%253A%252F%252Farxiv.org%252Fabs%252F2112.00861%2523anthropic.html

  295. Wikipedia Bibliography:

    1. ​ Claude (language Model)

    2. ​ Steven Levy

    3. ​ Quoc V. Le

    4. ​ Bo Li

    5. ​ Radha Poovendran

    6. ​ Holden Karnofsky

    7. ​ Paul Christiano

    8. ​ Jeremy Bernstein

    9. ​ Eliezer Yudkowsky

    10. ​ Will Knight

    11. ​ Danny Hernandez

    12. ​ Scott Johnston

    13. ​ Dario Amodei