Artificial intelligence
Models, labs, research and products shaping how AI is built, deployed and governed.
- WikiAI controlAI control designs safeguards that hold even if an AI model tries to subvert them: monitoring, trusted editing and sandboxing.Updated
- WikiAlphaFoldAlphaFold is Google DeepMind's AI system for predicting protein and biomolecular structures. Versions, accuracy, database and limits.Updated
- WikiAMDAMD designs CPUs, GPUs and AI accelerators for data centers, PCs and embedded systems. Its history, products, finances and recent deals as of October 2026.Updated
- WikiAnthropicAnthropic is the US AI company that makes the Claude models. Its structure, funding, IPO filing and 2026 Pentagon dispute.Updated
- WikiARC-AGIARC-AGI is a benchmark series of tasks easy for people and hard for AI. ARC-AGI-3 went from 0.51% to 62.7% for AI within six months.Updated
- WikiCAISI / CAISSI: the US government's AI testing centre at NISTThe US centre at NIST that tests frontier AI models, from AI Safety Institute to CAISI (2025) and CAISSI (2026).Updated
- WikiClaude MythosClaude Mythos is Anthropic's most capable model class, first released as a gated preview for cyber defence and later as Claude Fable.Updated
- WikiConstitutional AIConstitutional AI trains models with written principles and AI feedback instead of only human labels. How it works and what changed in 2026.Updated
- WikiDeepSeek-R1DeepSeek-R1 reported that reinforcement learning alone can teach a language model to reason. Its findings, publication and successors.Updated
- WikiECMWF AIFS (Artificial Intelligence Forecasting System)The AIFS is ECMWF's operational AI weather model, running since 2025 beside its physics-based IFS. Performance, energy use and limits.Updated
- WikiFigure AIFigure AI builds general-purpose humanoid robots and its own Helix AI model. Its robots, BMW deployments, funding and 2026 milestones.Updated
- WikiFrontier safety frameworks (responsible scaling policies)Frontier safety frameworks are AI companies' if-then rules for dangerous capabilities. How they work, who has one, and 2026 changes.Updated
- WikiGemini RoboticsGemini Robotics is Google DeepMind's family of robot AI models. What it does, how it evolved from RT-2 to Gemini Robotics 2, and its limits.Updated
- WikiGenesis Mission (US AI-for-science initiative)The Genesis Mission is the US government's DOE-led programme to accelerate science with AI, launched in November 2025. Goals and awards.Updated
- WikiGNoME (Graph Networks for Materials Exploration)GNoME is Google DeepMind's deep-learning project that predicted 2.2 million new crystal structures. What it found and why critics push back.Updated
- WikiGoogle DeepMindGoogle DeepMind is Alphabet's AI research lab, maker of Gemini and AlphaFold. Its origins, leadership, models and 2026 news.Updated
- WikiIntelIntel designs and manufactures processors and runs a contract foundry. Its history, 18A process, finances and new investors as of October 2026.Updated
- WikiNVIDIA Isaac GR00TIsaac GR00T is NVIDIA's platform of open humanoid robot foundation models, simulation tools and reference hardware. Versions, data and partners.Updated
- WikiIsomorphic LabsIsomorphic Labs is the Alphabet-backed AI drug discovery company led by Demis Hassabis. Funding, partners and clinical timeline.Updated
- WikiMETRMETR is an AI evaluation group known for measuring how long a task AI agents can complete, and for trials of AI's effect on developers.Updated
- WikiModel Context Protocol (MCP)The Model Context Protocol is an open standard for connecting AI apps and agents to tools and data, now run by the Linux Foundation.Updated
- WikiNVIDIANVIDIA designs GPUs, CPUs and networking for AI data centers. Its history, products, finances and recent developments as of October 2026.Updated
- WikiOpen X-EmbodimentOpen X-Embodiment is a pooled dataset of 1M+ robot trajectories from 22 robot types and the RT-X models trained on it. Why it mattered.Updated
- WikiOpenAIOpenAI is the US AI company behind ChatGPT and the GPT models. Its structure, funding, products and what changed in 2026.Updated
- WikiOpenAI o1OpenAI o1 is a reasoning model series trained with reinforcement learning to think in a chain of thought before answering.Updated
- WikiPhysical IntelligencePhysical Intelligence (π) is a startup building general-purpose robot foundation models such as π0 and π0.5. Its models, data approach and funding.Updated
- WikiRFdiffusionRFdiffusion is an open-source generative AI model from the Institute for Protein Design that designs new proteins and binders.Updated
- WikiReinforcement learning from human feedback (RLHF)RLHF trains AI models on human preference rankings. Where it came from, why it became the standard, and its limits.Updated
- WikiSamsung ElectronicsSamsung Electronics is a South Korean maker of memory chips, foundry services and devices. Its structure, chip business and 2026 results.Updated
- WikiScalable oversightScalable oversight asks how humans can supervise AI that outperforms them. Debate, AI-assisted judging and weak-to-strong research.Updated
- WikiSelf-driving labs (autonomous laboratories)Self-driving labs pair AI with robots that plan, run and learn from experiments. The A-Lab, the controversy, and the 2025–26 start-up wave.Updated
- WikiSparse autoencoders (interpretability)Sparse autoencoders split an AI model's internal activity into readable features. Their rise, scale-up and limits.Updated
- WikiSWE-benchSWE-bench tests whether AI can fix real GitHub issues. How it works, its Verified subset, and how scores rose from 2% to near 100%.Updated
- WikiTesla OptimusOptimus is Tesla's humanoid robot. Its history since 2022, the Fremont and Texas factory plans, repeated delays and status as of October 2026.Updated
- WikiTest-time computeTest-time compute is the computing power an AI model spends while answering. Spending more of it is how reasoning models improve.Updated
- WikiTSMC (Taiwan Semiconductor Manufacturing Company)TSMC is the Taiwanese contract chipmaker that builds chips designed by other companies. Its history, scale, process technology and 2026 results.Updated
- WikiUK AI Security Institute (AISI)The UK AI Security Institute tests frontier AI models for national-security risks. Its history, tools and key findings.Updated
- WikiUnitree RoboticsUnitree Robotics is a Hangzhou maker of quadruped and humanoid robots. Its shipments, 2026 Shanghai IPO, the US import ban, and role in research.Updated
- WikiWaymoWaymo runs driverless robotaxis in US cities. Its scale, safety data and AI approach, and why self-driving cars matter for embodied AI.Updated
- WikiWeatherNext (GraphCast, GenCast)Google DeepMind's AI weather models, from GraphCast and GenCast to WeatherNext 2: how they work and how they compare.Updated
- ● DevelopingAI for science tracker: milestones and what to watchLive tracker of AI for science: protein AI, AI weather models, materials, autonomous labs and AI drug discovery, with dated milestones.Updated 25 confirmed1 disputed
- ● DevelopingAI safety tracker: alignment, interpretability and evals in 2026Live tracker of AI safety milestones: interpretability results, evaluations, safety frameworks, incidents and institutes.Updated 29 confirmed
- ● DevelopingFrontier AI tracker: reasoning models and agentsA dated timeline of frontier AI milestones in reasoning models and AI agents, from o1 and DeepSeek-R1 to GPT-6, Claude Mythos and Gemini 4.Updated 34 confirmed
- ● DevelopingRobotics and embodied AI tracker: milestones to watchA running timeline of milestones in robot foundation models, humanoids and self-driving since the 2023 start of the VLA era, updated as of October 2026.Updated 33 confirmed
- ● DevelopingUS export controls on advanced AI chipsHow US rules on exporting advanced AI chips to China have changed since October 2022, who they affect, and what is still unresolved.Updated 16 confirmed
- AnalysisHow fast are AI agents really improving?AI agents' task horizons are doubling every few months on benchmarks, but real-world gains are harder to measure. The evidence, weighed.Updated
- AnalysisWhy AI chip supply is so concentratedWhy a few firms make the chips, memory and tools behind AI, what the evidence shows as of 2026, and competing views on change.Updated
- AnalysisDo AI-designed drugs work? What the clinical evidence showsAI-designed drug candidates are reaching patients. What trial data show so far, what they don't, and why "AI-designed" is a contested label.Updated
- AnalysisHas AI discovered new materials? The debate over GNoME and A-LabAI has predicted millions of crystals and robots have tried to make them. Why chemists dispute the claims, and what would count as proof.Updated
- AnalysisCan AI safety keep pace with AI capabilities?The central debate in AI safety in 2026: are evaluations, interpretability and oversight keeping up with fast-rising capabilities?Updated
- AnalysisShould games use AI-generated voices? The AI voice NPC debateAI voice characters promise open-ended conversation in games; performers want consent and pay. Fortnite's AI Darth Vader, the SAG-AFTRA deal and NVIDIA ACE.Updated
- AnalysisAre humanoid robots ready for real work? The 2026 debateEvidence on whether humanoid robots and robot foundation models are ready for factories and homes, with competing views on data, demand and safety.Updated
- AnalysisDo reasoning models really reason? The debate over their limitsReasoning models win maths olympiads yet fail some simple tasks, and their written reasoning is not always faithful. The evidence, weighed.Updated
- ExplainerAI for science in 2026: a crash courseA crash course on AI for science: protein prediction and design, AI weather models, materials and drug discovery, autonomous labs.Updated
- ExplainerAI safety and alignment in 2026: a crash courseA sourced crash course on AI safety: alignment, interpretability, evaluations, oversight, safety institutes and the 2026 frontier.Updated
- ExplainerFrontier AI in 2026: a crash courseA crash course on frontier AI in 2026: how reasoning models and AI agents work, who builds them, and where the frontier stands now.Updated
- ExplainerHow AI agents workWhat an AI agent is and how it works: language models using tools in a loop, computer use, MCP connectors and long tasks.Updated
- ExplainerHow frontier AI capabilities are measuredHow researchers measure what frontier AI can do: benchmarks like SWE-bench, Humanity's Last Exam and ARC-AGI, and METR time horizons.Updated
- ExplainerHow AI designs new proteins, drugs and materialsHow generative AI proposes new proteins, drug molecules and crystals, and why every design still has to be made and tested in a lab.Updated
- ExplainerHow AI predicts protein structuresWhy a protein's 3D shape matters, how AlphaFold-style models predict it from sequence, and what they still get wrong.Updated
- ExplainerHow AI safety testing works: evals, red teams and thresholdsHow frontier AI models are tested before release: dangerous-capability evals, jailbreak red-teaming, and why testing got harder.Updated
- ExplainerHow AI weather models workHow machine-learning weather models learn from decades of past weather, why they are so fast, and how forecasters now use them.Updated
- ExplainerHow large language models workA plain guide to large language models: the Transformer, scaling laws and human-feedback training, at beginner and expert level.Updated
- ExplainerHow mechanistic interpretability works: looking inside AIHow researchers reverse-engineer AI models: features, sparse autoencoders, circuit tracing and the 2026 Jacobian lens.Updated
- ExplainerHow reasoning models workHow AI reasoning models think step by step: chain of thought, reinforcement learning on checkable tasks and test-time compute.Updated
- ExplainerHow robots learn: demonstrations, trial and error, and dataA plain guide to how modern robots learn skills from human demonstrations, reinforcement learning and pooled datasets, and why data is the bottleneck.Updated
- ExplainerRobotics and embodied AI in 2026: a crash courseA crash course on robotics and embodied AI as of October 2026: how robots learn, robot foundation models, humanoids, self-driving and the open debates.Updated
- ExplainerSim-to-real: training robots in simulation for the real worldHow robots are trained in simulated worlds and then moved to real hardware, what the "reality gap" is, and how domain randomization bridges it.Updated
- ExplainerVision-language-action models: how robot foundation models workWhat vision-language-action (VLA) models are, how they turn camera images and instructions into robot motion, and the main models as of 2026.Updated
- ExplainerVoice gaming explained: voice control, voice chat and AI voicesHow games use your voice in 2026: spell-casting and ghost-talking games, proximity chat, PS5 voice and accessibility tools, and AI-voiced characters.Updated
- ExplainerWhat is AI alignment? Making AI do what we intendAI alignment explained: how labs train models to follow human intent, why it can fail, and what 2024–2026 studies found.Updated