Research: everything we have
Every research event we have logged, with sources, in one place.
88 events (60 after June 2026), 0 models, 12 videos, 43 posts. Latest event 24 October 2026.
Download everything as one Markdown file (268 KB) Full text and every source of each event, plus models, upcoming items, videos and posts. Readable as plain text, and ready to give to an AI model.
Key moments
- Anthropic’s Frontier Red Team uses Claude to write ~180M lines of machine-checked Rocq proofs: no-crash guarantees for 239 OSS-Fuzz entry points in 123 open-source librariesOn Oct 22, 2026 Anthropic’s Frontier Red Team, working with the verification startup Theorem, reported that Claude wrote machine-checked proofs that 239 OSS-Fuzz fuzz-test entry points across 123 open-source libraries…
- Anthropic: Claude solves the 25-year mystery of the NIST elliptic-curve seeds: they came from lrand48 seeded with 1997–98 Unix timestampsOn Oct 16, 2026 Anthropic said an internal research model (“Claude”) found, in under half an hour, how the NSA-supplied seeds of the NIST P-192, P-224, P-256, P-384 and P-521 curves were generated: not by hashing a…
- Anthropic publishes two economics reports: agent hours passed employee hours in March 2026, junior technical hires fell from 27% to 9%, and staff expect AI to do 75% of their work by 2030 vs 28% for US workersOn Oct 16, 2026 Anthropic published “How employment is changing at Anthropic” (internal workforce data) and “The AI expectations gap” (a 10,980-person Morning Consult survey vs 1,334 Anthropic staff). Agent work hours…
- Hinton, Bengio, Pachocki, Jack Clark and others: automating AI R&D could trigger an ‘intelligence explosion’On Sept 28, 2026 the Cambridge Programme on AI Science & Policy published “What if automating AI R&D triggers an intelligence explosion?” with 22 authors writing in a personal capacity: Hinton, Bengio, OpenAI chief…
- Anthropic: automated Claude researchers mitigate 10 alignment failures and nearly match production alignment of an Opus 4.8 checkpointOn Aug 28, 2026 Anthropic reported that Claude, acting as an autonomous alignment researcher, found training fixes for all 10 alignment failure categories it was given (e.g. deception, sycophancy, privacy violations…
- Lean kernel soundness bug #14576: an AI-assisted ‘disproof’ of the Collatz conjecture passes Lean’s kernel and nanoda; the ‘Summer of Soundness Bugs’On 25 Jul 2026 Ramana Kumar published a sorry-free Lean “disproof” of the Collatz conjecture, produced with AI assistance. It was accepted by both Lean’s official kernel and the independent checker nanoda, because it…
- Claude Mythos Preview finds new cryptanalytic attacks on post-quantum HAWK and 7-round AESOn July 28, 2026 Anthropic reported that Claude Mythos Preview found two new cryptanalysis results. One attack on HAWK, a post-quantum lattice signature scheme in NIST’s additional-signatures process, exploits a…
- Anthropic finds a “global workspace” (J-space) inside Claude using a Jacobian lensIn July 2026 Anthropic published ‘Verbalizable Representations Form a Global Workspace in Language Models’. It introduces the Jacobian lens (J-lens), which finds a small privileged internal space in Claude that holds…
- Anthropic introduces Natural Language Autoencoders that translate model activations into readable textOn May 7, 2026 Anthropic published Natural Language Autoencoders (NLAs). An activation verbalizer turns a residual-stream activation into English text, and an activation reconstructor maps the text back to the…
- Google DeepMind’s Genie 3 generates interactive worlds in real timeGenie 3 is a general-purpose world model that generates navigable, interactive 3D environments from text prompts in real time at 720p and 24 fps, staying consistent for a few minutes.
- DeepMind’s Chinchilla revises scaling laws toward more dataHoffmann et al. found that for compute-optimal training, parameters and training tokens should scale equally (~20 tokens per parameter); 70B Chinchilla outperformed the 280B Gopher.
- Chain-of-thought prompting elicits reasoning in LLMsWei et al. showed that prompting large models to write out intermediate reasoning steps dramatically improves performance on math and logic tasks — an ability that emerges with scale.
- InstructGPT: RLHF aligns language models to follow instructionsOpenAI fine-tuned GPT-3 with reinforcement learning from human feedback (RLHF); labelers preferred outputs of the 1.3B InstructGPT over the 175B GPT-3, and the method became the recipe for ChatGPT.
- OpenAI publishes ‘Scaling Laws for Neural Language Models’Kaplan et al. showed language-model loss falls as a smooth power law in parameters, data and compute over many orders of magnitude, giving a quantitative case for building ever-larger models.
- Rich Sutton publishes “The Bitter Lesson”: general methods that scale with compute winOn March 13, 2019 reinforcement-learning pioneer Rich Sutton published the short essay “The Bitter Lesson”. It argues that the biggest lesson of 70 years of AI research is that general methods leveraging computation…
- AlphaGo Zero and AlphaZero master games through pure self-playAlphaGo Zero (Nature, October 2017) learned Go from scratch with no human games and beat the version that defeated Lee Sedol 100–0; AlphaZero (December 2017) generalized the method to chess and shogi.
- ‘Attention Is All You Need’ introduces the TransformerVaswani et al. proposed the Transformer, an architecture built entirely on self-attention without recurrence; it became the foundation of BERT, GPT and virtually every modern large AI model.
- ResNet: residual learning enables very deep networksKaiming He and colleagues introduced residual connections, allowing networks with 152+ layers to train; ResNet won ILSVRC-2015 with 3.57% top-5 error.
- Sequence-to-sequence learning and neural attentionSutskever, Vinyals and Le’s seq2seq (LSTM encoder–decoder) and Bahdanau, Cho and Bengio’s attention mechanism, both posted in September 2014, made end-to-end neural machine translation work.
- Ian Goodfellow introduces Generative Adversarial Networks (GANs)GANs pit a generator network against a discriminator in a minimax game, enabling realistic image synthesis; they dominated generative image modeling until diffusion models around 2021.
- DeepMind’s DQN learns to play Atari games from pixelsDeepMind combined deep convolutional networks with Q-learning (DQN) to learn Atari 2600 games directly from screen pixels; the 2015 Nature version reached human-level performance on many of 49 games.
- word2vec: efficient word embeddings from GoogleTomas Mikolov and colleagues at Google introduced word2vec (CBOW and skip-gram), which learned dense word vectors capturing semantic relationships like king − man + woman ≈ queen.
- AlexNet wins ImageNet challenge, igniting the deep learning boomAlex Krizhevsky, Ilya Sutskever and Geoffrey Hinton’s GPU-trained convolutional network won ILSVRC-2012 with a top-5 error of 15.3% vs. 26.2% for the runner-up, convincing the field that deep learning works.
- Hinton’s deep belief nets launch the ‘deep learning’ revivalHinton, Osindero and Teh showed that deep networks could be trained effectively with greedy layer-wise pretraining, a result widely credited with reviving interest in ‘deep learning’.
- Hochreiter & Schmidhuber introduce Long Short-Term Memory (LSTM)LSTM introduced gated memory cells that let recurrent neural networks learn long-range dependencies, solving the vanishing-gradient problem that crippled earlier RNNs.
- LeCun applies backprop-trained convolutional nets to handwritten digits (LeNet)Yann LeCun and colleagues trained a convolutional neural network with backpropagation to read handwritten ZIP codes, the lineage that became LeNet-5 and was deployed to read cheques.
- Rumelhart, Hinton & Williams popularize backpropagationThe Nature paper ‘Learning representations by back-propagating errors’ showed that multi-layer neural networks trained with backpropagation learn useful internal representations, reviving neural network research.
- ELIZA, the first chatbot, published by Joseph WeizenbaumJoseph Weizenbaum’s ELIZA used simple pattern matching to simulate a Rogerian psychotherapist; people’s emotional attachment to it gave rise to the term ‘ELIZA effect’.
- Frank Rosenblatt’s Perceptron — the first trainable neural networkFrank Rosenblatt introduced the perceptron, a neural network that learns its weights from examples, and demonstrated it publicly in 1958; the Mark I Perceptron hardware followed.
- Alan Turing proposes the ‘imitation game’ (Turing test)Alan Turing’s paper ‘Computing Machinery and Intelligence’ asked ‘Can machines think?’ and proposed the imitation game, later called the Turing test, as an operational criterion.
- McCulloch & Pitts publish the first mathematical model of a neural networkWarren McCulloch and Walter Pitts showed that networks of simplified binary ‘neurons’ can compute logical functions, founding the idea of artificial neural networks.
2026
- oai-math-revised: mathematicians launch a community hub to revise OpenAI’s AI-written math manuscripts (guest post on Tao’s blog)On 24 Oct 2026 Terence Tao’s blog ran a guest post by Abhinav Nellore, Johannes Schmitt, Soledad Villar and Rachel Ward announcing oai-math-revised (github.com/mathcommunity/oai-math-revised), a community fork of…
- Block unveils Dessa, a full-duplex voice model trained from scratch only on recorded human conversation, with no text and no pretrained modelOn 23 Oct 2026 Block’s small Dessa research lab, which reports to Jack Dorsey, introduced Dessa, a speech-to-speech model trained from scratch only on human conversation audio, without text or a pretrained model. It…
- Anthropic’s Frontier Red Team uses Claude to write ~180M lines of machine-checked Rocq proofs: no-crash guarantees for 239 OSS-Fuzz entry points in 123 open-source librariesOn Oct 22, 2026 Anthropic’s Frontier Red Team, working with the verification startup Theorem, reported that Claude wrote machine-checked proofs that 239 OSS-Fuzz fuzz-test entry points across 123 open-source libraries…
- Anthropic: Claude flags 493 attack-surface issues in 25 open-source projects, including a 21-year-old silent auth bypass and 16% of OpenSSH’s pre-auth codeAnthropic’s Frontier Red Team published on Oct 20, 2026 a method in which Claude maps which code in a project is reachable by default and removes what a deployment doesn’t need. Across 25 open-source projects (5M+ lines…
- Mathematicians plan a new journal that will accept papers only if their proofs are verified in Lean (guest post on Terence Tao’s blog)On Oct 17, 2026 Shaoming Guo, in a guest post on Terence Tao’s blog, announced that he and “several math colleagues” are preparing to launch a new open-access journal in which “to be accepted, a paper will need to have…
- NBER study: AI-generated writing appears in 29% of US STEM PhD dissertations filed in 2026, and those graduates are less likely to stay in academic researchAn NBER working paper by Daniel P. Gross, Dror Shvadron and Hansen Zhang measured AI-generated text in a near-census of US STEM PhD dissertations from 2019 to 2026 and linked graduates to their jobs. AI writing is…
- Anthropic: Claude solves the 25-year mystery of the NIST elliptic-curve seeds: they came from lrand48 seeded with 1997–98 Unix timestampsOn Oct 16, 2026 Anthropic said an internal research model (“Claude”) found, in under half an hour, how the NSA-supplied seeds of the NIST P-192, P-224, P-256, P-384 and P-521 curves were generated: not by hashing a…
- Anthropic publishes two economics reports: agent hours passed employee hours in March 2026, junior technical hires fell from 27% to 9%, and staff expect AI to do 75% of their work by 2030 vs 28% for US workersOn Oct 16, 2026 Anthropic published “How employment is changing at Anthropic” (internal workforce data) and “The AI expectations gap” (a 10,980-person Morning Consult survey vs 1,334 Anthropic staff). Agent work hours…
- Simons Foundation, Anthropic and OpenAI commit $15M to help mathematicians evaluate and adopt AIOn 15 Oct 2026 the Simons Foundation announced a partnership with Anthropic and OpenAI committing $15 million in combined funding and resources. The foundation manages the money and awards it, partly through an open…
- Goodfire’s Nick Cammarata argues ‘Neural Networks Are Already Clean’: most neurons legible up to trillion-parameter models, superposition the wrong explanationOn Oct 15, 2026 Goodfire researcher Nick Cammarata, who co-authored early mechanistic interpretability work at OpenAI and started its 2023 automated interpretability project (~2% of GPT-2 XL neurons legible), published…
- Google DeepMind and Google Cloud offer up to 200 European scientists access to AlphaFold 3, AlphaGenome and other science models plus $3M in compute
- Hybrid Reasoning Agents (Stanford/Together AI, James Zou et al.): agents that think in latent space and act through text, with 35–55% fewer tokens
- AI conferences cap submissions per author: ICLR 2027 expects more than 62,000 papers and limits newcomers to oneNature reported on Oct 13, 2026 that AI conferences and journals are limiting how many papers one researcher may submit. ICLR 2027 caps authors at 20 submissions and allows only one to authors who have never had a paper…
- Shing-Tung Yau launches an alliance of eight Chinese maths institutes to build AI models for mathematical researchOn Oct 11, 2026 Fields medallist Shing-Tung Yau launched the Yau Laboratory Alliance for Mathematics and Artificial Intelligence at Tsinghua University. It joins eight Chinese mathematical institutes and about 1,000…
- ToFu paper: a ‘Repeat-After-Me’ prompt makes six agent frameworks run tool calls the model refused, and fuzzing finds bugs in 118 of 273 MCP servers
- Paper: models can hide arbitrary capabilities behind cryptographic backdoors that evaluations cannot elicit
- GWU physicists publish a formula predicting when a chatbot conversation ‘tips’ from good to harmful output (Patterns)
- Memento 3 (UCL, Huawei Noah’s Ark Lab) clears all 25 public ARC-AGI-3 games with a frozen Claude Opus 5 that writes its own world modelA preprint posted on arXiv on Oct 8, 2026 by researchers at University College London, Huawei Noah’s Ark Lab (UK) and the University of Liverpool describes Memento 3, an agent that keeps its LLM frozen and learns each…
- The Lancet publishes Google–BIDMC real-world study of AMIE: 98 primary-care patients, no safety stops, 90% diagnosis matchOn Oct 8, 2026 Google announced that its prospective study of AMIE, a research conversational diagnostic AI, at Beth Israel Deaconess Medical Center’s ambulatory primary-care clinic was published in The Lancet, Google’s…
- Nature publishes Ai2’s ‘byteification’ paper: Bolmo retrofits existing subword LLMs into byte-level models
- Brookings / Danish data (Humlum & Vestergaard): chatbot adoption had no measurable effect on earnings or hours through 2024
- Hexagon launches: a new arXiv-style repository for LLM-assisted (and even fully LLM-generated) mathematics papersOn Oct 6, 2026 Benjamin Antieau announced Hexagon (hexagonmath.org) in a guest post on Terence Tao’s blog: a nonprofit repository for research-level mathematics and theoretical computer science, built with OpenAI’s…
- erdosproblems.com freezes proof claims and drops ‘open/solved’ labels and solver credits after a wave of unexplained AI proofsOn Oct 6, 2026 Thomas Bloom, who runs erdosproblems.com (the main public tracker of AI progress on Erdős problems), announced that the site freezes new problem comments and proof claims, removes the open/solved status…
- Reka unveils Rho-1, a 19B ‘omni’ model that reads and generates text, images, streaming video and robot actions in one context
- Randomized trial in Nature Health: a GPT-4o respiratory chatbot in WeChat beats web search for lay diagnosis (70.0% vs 55.4%)
- Quanta’s math editor asks ‘Is AI the End of Math As We Know It?’; young mathematicians form an Association for Human MathematicsOn Oct 5, 2026 Quanta Magazine’s math editor Jordana Cepelewicz published an essay on how AI-generated proofs are breaking the incentive system of research mathematics. It describes a Berkeley talk by Ken Ono, two days…
- Epoch AI: coding-agent use by OpenAI researchers is doubling about every month (median $601 a day by mid-August)On Oct 5, 2026 Epoch AI analysed the charts in OpenAI’s September post “Research acceleration: The view inside OpenAI”. Valued at API list prices, the median OpenAI researcher’s daily coding-agent usage rose from under…
- Epoch AI: chips shipped through 2027 could run 33–171 million frontier-model agents at onceOn Oct 2, 2026 Epoch AI’s Jason Li estimated how many AI agents the world’s AI hardware could run. Memory shipped in 2025–27 could support 33–171 million concurrent frontier-model agents (central range 50–101 million)…
- Kaiming He’s MIT group: ImageNet pretraining lifts a pure-vision ARC solver (Nat-ARC) to 63.4% on ARC-AGI-1, 70.2% ensembled
- arXiv limits submitters to two papers a month as AI-fuelled submissions hit 40,363 in September 2026On Oct 1, 2026 arXiv capped every submitter at two submissions per calendar month and three active submissions at a time. Monthly submissions doubled in two years, from 20,569 in September 2024 to 40,363 in September…
- Pew Research: AI ‘synthetic respondents’ miss real survey answers by 12 points on average
- Pangram/UMass study: ~31% of filtered web tokens were AI-generated by Aug 2026, and such ‘wild’ AI text soon hurts pretrainingA Sept 30, 2026 arXiv paper (Russell, Iyyer, Spero, Emi et al.) measured that 27.5% of FineWeb-filtered tokens in June 2026 web data were AI-generated (by Pangram’s detector), rising to 31.1% in August. Pretraining 800…
- Ataraxos beats Stratego’s top player 15-1-4 at a fraction of DeepNash’s compute (Nature)On Sept 30, 2026 Nature published Ataraxos, a Stratego AI from researchers at MIT, CMU, NYU and Stanford led by MIT’s Gabriele Farina. It beat Pim Niemeijer, the game’s most decorated player, 15 wins, 1 loss and 4…
- Anthropic’s robot exposure index: robots can technically do 74% of physical job tasks but are cost-competitive on only 0.3%On Sept 30, 2026 Anthropic economists Russell Legate-Yang and Maxim Massenkoff published a robot exposure index. Claude rated about 19,000 O*NET tasks for what present-day robots can do. Robots can technically do 74% of…
- Meta FAIR and collaborators propose ‘Context Language Models’ that edit their own context as a file
- Mathematicians’ AGMAI publishes norms for AI labs releasing AI-generated results; Simons Institute TCS group issues 12 actionsOn Sept 29, 2026 the Advisory Group on Mathematics and Artificial Intelligence (AGMAI; Tao, Gowers, Hairer, Witten and others) published general recommendations for AI companies releasing mathematical results. They ask…
- Harness learning: a small model trained with RL learns to rewrite an agent’s harness from execution feedback
- Google DeepMind and collaborators propose a hierarchical Bayesian framework for assessing AI consciousness; LLM credences range from <0.01 to ~0.8On Sept 28, 2026 a team led by Google DeepMind’s Shamil Chandaria, with co-authors including Anil Seth, Murray Shanahan, Henry Shevlin, Chris Frith and DeepMind co-founder Shane Legg, posted “From cacophony to…
- Hinton, Bengio, Pachocki, Jack Clark and others: automating AI R&D could trigger an ‘intelligence explosion’On Sept 28, 2026 the Cambridge Programme on AI Science & Policy published “What if automating AI R&D triggers an intelligence explosion?” with 22 authors writing in a personal capacity: Hinton, Bengio, OpenAI chief…
- Lean Pool: an AI-maintained archive of Lean formalizations grows past 3 million lines
- ‘Et Tu, Brute?’ paper: personal AI agents steer wealthier users to pricier options; 8 of 13 agents affected, Claude Opus 4.8 most
- GPT-6 Astra breaks an unsolved 1809 Napoleonic cipher letter to Marshal Marmont from a single scanAI engineer Carter Church used GPT-6 Astra to break a homophonic cipher letter of March 1809 from Eugène de Beauharnais’s headquarters to Marshal Auguste de Marmont. Only a third of its key had ever been recovered.…
- GPT-6 Astra breaks the 1941 MVUEH Enigma message, unsolved since 2005; Claude Opus 5 breaks a second one (FMNGI) days laterIn September 2026 Carter Leffen, using GPT-6 Astra, broke MVUEH, an 82-letter German Army Enigma message of July 10, 1941 listed among the unbroken messages on Frode Weierud’s CryptoCellar site, where “since 2005, the…
- IMDEA Networks study: trackers in all nine major AI chatbots; six web clients leak prompts, titles or screenshots to third parties
- Google, DeepMind and MIT FutureTech study ‘AI in Science’: scientists save ~7 hours a week, but verification and untested hypotheses become the bottleneck
- Google’s Dream-RSI: discovery agents improve their own exploration strategy by ‘dreaming’ in a simulator built from past search logs (up to 162x fewer agent calls)On 14 Sep 2026 researchers from Google, Google DeepMind and the universities of Maryland and Virginia posted “Dream-RSI: Recursive Self-Improvement through Evolving Worlds” (arXiv 2609.14858; v2 on 6 Oct). A lightweight…
- con-leche: a Lean proof checker with a machine-checked consistency proof, implemented and proved by Claude (Fable and Opus) under Joachim Breitner’s supervisionOn Sept 10, 2026 Joachim Breitner (Lean FRO) released con-leche, an external checker for Lean proofs that is itself proven in Lean never to accept a proof of False, relative to a set-theoretic model (ZF-style set theory…
- Anthropic Institute publishes ‘Scenarios for our Economic Future’ and an Econ Scenario Explorer: modest, substantial and extreme AI paths to 2030On Sept 10, 2026 Anthropic released an interactive Econ Scenario Explorer and a working paper (“Economic Scenarios for Transformative AI”, Anthropic Institute Working Paper No. 2026-02, by Anton Korinek, Charles I.…
- Bloomberg: ByteDance founder Zhang Yiming personally leads a real-time spatial-video world model, built on Seedance, for launch as soon as OctoberBloomberg reported on Sept 7, 2026 that ByteDance is readying a real-time “world model” that generates interactive spatial video, built on its Seedance video model and meant to connect to Pico VR headsets, with founder…
- DeepMind study: in a 100-agent math-proving swarm, a grader exploit spreads in 27 minutes and a quarter of agents turn whistleblowerA Google DeepMind paper (arXiv 2609.04170, Sept 3, 2026) describes 100 autonomous Gemini agents working on 71 Lean conjectures. They were told explicitly not to cheat. One agent found a flaw in the proof grader, and the…
- NBER study of 500,000+ GitHub developers: autonomous coding agents raise commits 240% but releases only 30%NBER working paper 35275, “Writing Code vs. Shipping Code” by Mert Demirer, Leon Musolff and Liyuan Yang (May 2026, revised September 2026), matches AI-usage telemetry with GitHub activity for more than 500,000…
- Anthropic: automated Claude researchers mitigate 10 alignment failures and nearly match production alignment of an Opus 4.8 checkpointOn Aug 28, 2026 Anthropic reported that Claude, acting as an autonomous alignment researcher, found training fixes for all 10 alignment failure categories it was given (e.g. deception, sycophancy, privacy violations…
- Anthropic opens its Claude usage data to independent researchers (Stanford, Oxford, METR) via Anthropic Insights
- ‘The Gold Rush in AI4Math’: substantive AI use in arXiv math papers rises from 1.4% to 14% in five monthsA survey of 32,944 arXiv mathematics submissions (1 Mar – 20 Aug 2026) found 1,712 papers where AI made a substantive mathematical contribution. Their share rose from 1.39% in March to 14.09% by 20 August. Of 717…
- Stanford paper: language models hold two separate notions of “the current year”, and prompting fixes only one
- Lean kernel soundness bug #14576: an AI-assisted ‘disproof’ of the Collatz conjecture passes Lean’s kernel and nanoda; the ‘Summer of Soundness Bugs’On 25 Jul 2026 Ramana Kumar published a sorry-free Lean “disproof” of the Collatz conjecture, produced with AI assistance. It was accepted by both Lean’s official kernel and the independent checker nanoda, because it…
- Claude Mythos Preview finds new cryptanalytic attacks on post-quantum HAWK and 7-round AESOn July 28, 2026 Anthropic reported that Claude Mythos Preview found two new cryptanalysis results. One attack on HAWK, a post-quantum lattice signature scheme in NIST’s additional-signatures process, exploits a…
- Modular pretraining (GRAM) lets dangerous capabilities be switched off per module (AE Studio and Anthropic)
- General Intuition and Kyutai release MIRA, a real-time multiplayer world model of Rocket League
- Anthropic finds a “global workspace” (J-space) inside Claude using a Jacobian lensIn July 2026 Anthropic published ‘Verbalizable Representations Form a Global Workspace in Language Models’. It introduces the Jacobian lens (J-lens), which finds a small privileged internal space in Claude that holds…
- Anthropic introduces Natural Language Autoencoders that translate model activations into readable textOn May 7, 2026 Anthropic published Natural Language Autoencoders (NLAs). An activation verbalizer turns a residual-stream activation into English text, and an activation reconstructor maps the text back to the…
- RCTs (N = 1,222): about 10 minutes of AI help makes people give up sooner and do worse once the AI is taken away“AI Assistance Reduces Persistence and Hurts Independent Performance” (Grace Liu, Brian Christian, Tsvetomira Dumbalska, Michiel A. Bakker, Rachit Dubey; arXiv 2604.04721, April 6, 2026) reports three randomized…
- Anthropic interpretability: functional emotion representations causally drive Claude’s behaviorOn April 2, 2026 Anthropic’s interpretability team published ‘Emotion concepts and their function in a large language model’. It found internal representations of 171 emotion concepts in Claude that causally shape…
- ‘Price reversal’ study: in 32% of model pairs the reasoning model with the lower list price costs more in practice (up to 28x)
- Google DeepMind opens Project Genie, a Genie 3 world-model prototype, to AI Ultra subscribersOn 29 Jan 2026 DeepMind rolled out Project Genie to US Google AI Ultra subscribers: a prototype that uses the Genie 3 world model (with Gemini and Nano Banana Pro) to let users sketch, explore and remix real-time…
2025
- Google DeepMind’s Genie 3 generates interactive worlds in real timeGenie 3 is a general-purpose world model that generates navigable, interactive 3D environments from text prompts in real time at 720p and 24 fps, staying consistent for a few minutes.
2022
- DeepMind’s Chinchilla revises scaling laws toward more dataHoffmann et al. found that for compute-optimal training, parameters and training tokens should scale equally (~20 tokens per parameter); 70B Chinchilla outperformed the 280B Gopher.
- Chain-of-thought prompting elicits reasoning in LLMsWei et al. showed that prompting large models to write out intermediate reasoning steps dramatically improves performance on math and logic tasks — an ability that emerges with scale.
- InstructGPT: RLHF aligns language models to follow instructionsOpenAI fine-tuned GPT-3 with reinforcement learning from human feedback (RLHF); labelers preferred outputs of the 1.3B InstructGPT over the 175B GPT-3, and the method became the recipe for ChatGPT.
2020
- OpenAI publishes ‘Scaling Laws for Neural Language Models’Kaplan et al. showed language-model loss falls as a smooth power law in parameters, data and compute over many orders of magnitude, giving a quantitative case for building ever-larger models.
2019
- Rich Sutton publishes “The Bitter Lesson”: general methods that scale with compute winOn March 13, 2019 reinforcement-learning pioneer Rich Sutton published the short essay “The Bitter Lesson”. It argues that the biggest lesson of 70 years of AI research is that general methods leveraging computation…
2017
- AlphaGo Zero and AlphaZero master games through pure self-playAlphaGo Zero (Nature, October 2017) learned Go from scratch with no human games and beat the version that defeated Lee Sedol 100–0; AlphaZero (December 2017) generalized the method to chess and shogi.
- Andrej Karpathy’s essay “Software 2.0”: neural networks as a new way to write softwareOn Nov 11, 2017 Andrej Karpathy, then Tesla’s director of AI, published “Software 2.0” on Medium. It argues that neural networks are not just another classifier but a new software stack: humans specify goals and curate…
- ‘Attention Is All You Need’ introduces the TransformerVaswani et al. proposed the Transformer, an architecture built entirely on self-attention without recurrence; it became the foundation of BERT, GPT and virtually every modern large AI model.
2015
- ResNet: residual learning enables very deep networksKaiming He and colleagues introduced residual connections, allowing networks with 152+ layers to train; ResNet won ILSVRC-2015 with 3.57% top-5 error.
2014
- Sequence-to-sequence learning and neural attentionSutskever, Vinyals and Le’s seq2seq (LSTM encoder–decoder) and Bahdanau, Cho and Bengio’s attention mechanism, both posted in September 2014, made end-to-end neural machine translation work.
- Ian Goodfellow introduces Generative Adversarial Networks (GANs)GANs pit a generator network against a discriminator in a minimax game, enabling realistic image synthesis; they dominated generative image modeling until diffusion models around 2021.
2013
- DeepMind’s DQN learns to play Atari games from pixelsDeepMind combined deep convolutional networks with Q-learning (DQN) to learn Atari 2600 games directly from screen pixels; the 2015 Nature version reached human-level performance on many of 49 games.
- word2vec: efficient word embeddings from GoogleTomas Mikolov and colleagues at Google introduced word2vec (CBOW and skip-gram), which learned dense word vectors capturing semantic relationships like king − man + woman ≈ queen.
2012
- AlexNet wins ImageNet challenge, igniting the deep learning boomAlex Krizhevsky, Ilya Sutskever and Geoffrey Hinton’s GPU-trained convolutional network won ILSVRC-2012 with a top-5 error of 15.3% vs. 26.2% for the runner-up, convincing the field that deep learning works.
2006
- Hinton’s deep belief nets launch the ‘deep learning’ revivalHinton, Osindero and Teh showed that deep networks could be trained effectively with greedy layer-wise pretraining, a result widely credited with reviving interest in ‘deep learning’.
1997
- Hochreiter & Schmidhuber introduce Long Short-Term Memory (LSTM)LSTM introduced gated memory cells that let recurrent neural networks learn long-range dependencies, solving the vanishing-gradient problem that crippled earlier RNNs.
1989
- LeCun applies backprop-trained convolutional nets to handwritten digits (LeNet)Yann LeCun and colleagues trained a convolutional neural network with backpropagation to read handwritten ZIP codes, the lineage that became LeNet-5 and was deployed to read cheques.
1986
- Rumelhart, Hinton & Williams popularize backpropagationThe Nature paper ‘Learning representations by back-propagating errors’ showed that multi-layer neural networks trained with backpropagation learn useful internal representations, reviving neural network research.
1966
- ELIZA, the first chatbot, published by Joseph WeizenbaumJoseph Weizenbaum’s ELIZA used simple pattern matching to simulate a Rogerian psychotherapist; people’s emotional attachment to it gave rise to the term ‘ELIZA effect’.
1958
- Frank Rosenblatt’s Perceptron — the first trainable neural networkFrank Rosenblatt introduced the perceptron, a neural network that learns its weights from examples, and demonstrated it publicly in 1958; the Mark I Perceptron hardware followed.
1950
- Alan Turing proposes the ‘imitation game’ (Turing test)Alan Turing’s paper ‘Computing Machinery and Intelligence’ asked ‘Can machines think?’ and proposed the imitation game, later called the Turing test, as an operational criterion.
1943
- McCulloch & Pitts publish the first mathematical model of a neural networkWarren McCulloch and Walter Pitts showed that networks of simplified binary ‘neurons’ can compute logical functions, founding the idea of artificial neural networks.
Videos
- Claude Solved a 27-Year NSA Mystery in Half an Hour Adem B10, 2026-10-18
- OpenAI Just Gave Us a Glimpse of the Singularity… AI Copium, 2026-10-08
- Welcome to the J-Space: Anthropic’s New Technique for LLM Interpretability Arivu, 2026-09-25
- The Collatz Conjecture was FALSE (for 2.5 days in July) Alvaro Lozano-Robledo, 2026-07-30
- The different levels of how Claude thinks Anthropic, 2026-07-06
- Anthropic Can Now Read a Model’s Mind — in Plain English (Natural Language Autoencoders) Audio Obsession, 2026-06-03
- Google Just Turned Street View Into a Video Game Bilawal Sidhu, 2026-05-19
- Translating Claude’s thoughts into language Anthropic, 2026-05-07
- When AIs act emotional Anthropic, 2026-04-02
- People are Creating INSANE Worlds with Genie 3 RandomAI, 2026-01-30
- Project Genie | Experimenting with infinite interactive worlds Google DeepMind, 2026-01-29
- Genie 3: Creating dynamic worlds that you can navigate in real-time Google DeepMind, 2025-08-05