A free encyclopedia of AI
The AI encyclopedia that checks the numbers.
Most claims about AI get repeated, not verified. Here they get traced to the filing, the trial or the court record, and where the evidence is thin the article says so. Every concept is also written five ways, from one plain sentence to the open research question.
- 1.6%of 1,524 cleared medical AI devices cite a trial
- 19% slowermeasured, while developers reported 20% faster
- 90% / 40%deflected versus actually resolved
- 1,313filings sanctioned for fabricated citations
Try it — drag to change the depth
Curious1 of 5
CuriousPracticalHands-on
TechnicalFrontier
Browse
TransformerTokenAI AgentDiffusion ModelSupervised LearningImage ClassificationAI AlignmentMachine LearningVector DatabaseSpeech RecognitionRecommender SystemEmbeddingsLarge Language Model (LLM)Tool UseGAN (Generative Adversarial Network)OverfittingObject DetectionBias & FairnessDeep LearningOpen-Weight ModelsText-to-SpeechTime Series ForecastingTransformerTokenAI AgentDiffusion ModelSupervised LearningImage ClassificationAI AlignmentMachine LearningVector DatabaseSpeech RecognitionRecommender SystemEmbeddingsLarge Language Model (LLM)Tool UseGAN (Generative Adversarial Network)OverfittingObject DetectionBias & FairnessDeep LearningOpen-Weight ModelsText-to-SpeechTime Series Forecasting
312
concepts
944
sources
2+
per entry
5
depths
Written to be checked, not trusted. Every entry lists its sources, and every source links to a lookup.
Where you left off
Last viewed on this deviceYour bookmarks
Saved on this deviceEleven fields of AI
See how they connectDeep Learning
37 conceptse.g. TransformerLanguage & LLMs
40 conceptse.g. TokenAI Agents
17 conceptse.g. AI AgentGenerative AI
12 conceptse.g. Diffusion ModelMachine Learning
47 conceptse.g. Supervised LearningComputer Vision
16 conceptse.g. Image ClassificationSafety & Ethics
34 conceptse.g. AI AlignmentFoundations
59 conceptse.g. Machine LearningTools & Ecosystem
16 conceptse.g. Vector DatabaseSpeech & Audio
13 conceptse.g. Speech RecognitionApplied AI
21 conceptse.g. Recommender SystemArtifact of the day
8 August 2026
Assuming retrieval fixes hallucination. It reduces it. The model can still ignore, misread, or blend the retrieved text.
Retrieval-Augmented Generation (RAG) · Language & LLMs
312connected concepts
See the whole map of AI
Every idea, every link between them, on one page. Drag it around and find where things fit.
Start with the big questions
The big questions everything else hangs off — answered properly, with every claim linked into the encyclopedia.
What is artificial intelligence?AI is the field of making machines do things that seem to require intelligence, and that definition has moved every single time the machines succeeded.Read the answer →What is machine learning?Machine learning is programming with examples instead of rules, an approach that took over because writing the rules down turned out to be impossible.Read the answer →What is AGI?AGI, artificial general intelligence, is a system with broad human-level capability across domains.Read the answer →
Writing on how AI behaves
All articles
Language & meaning
How transformers work: the architecture that ate AI
One 2017 paper replaced the entire previous approach to sequence modeling and made modern AI possible. Here's what a…
June 26, 2026
Building with AI
How AI models are trained: from raw text to a system that helps
A language model isn't programmed, it's grown, in stages, from a firehose of text into a system that answers…
July 12, 2026
Evaluation & evidence
The EU delayed the part with no standards
Three AI Act obligations take effect today and the widely reported headline says the opposite. What was deferred,…
August 2, 2026
Evaluation & evidence
72 seconds, or 30 minutes, and both are trials
Territory 11 opens on the first subject this corpus has examined where the evidence is genuinely good. Registered…
August 2, 2026
Evaluation & evidence
The 12% was non-inferior, and P was 0.41
MASAI is the best-evidenced AI deployment in medicine and the headline everyone quoted describes a result the trial…
August 2, 2026
Foundations
Nine subjects, and the model was almost never the blocker
Territory 10 closes. Across nine enterprise deployment subjects the binding constraint was organisational in every…
August 2, 2026
Stay in the loop
New concepts, in plain English.
We're adding entries every week — the terms you keep running into, explained without the jargon. Drop your email and we'll send the good ones. No spam, unsubscribe anytime.
All 312 concepts, A to Z
Explore the mapDeep Learning
37TransformerThe neural-network architecture behind almost every modern AI model — built around attention, which lets it weigh every word against every other, all at once.EmbeddingsTurning words (or images, or anything) into lists of numbers, arranged so that similar meanings end up close together.AttentionThe mechanism that lets an AI decide which other words matter when interpreting each word — the core idea behind transformers.Neural NetworkA system of simple connected units that learns patterns from examples — the foundation underneath deep learning and modern AI.BackpropagationThe algorithm that works out which weights caused a mistake and by how much — the reason neural networks can learn at all.Gradient DescentWalking downhill on the error surface, one small step at a time — how a model's weights actually get updated.
View all 37 in Deep LearningLanguage & LLMs
40TokenThe small piece of text an AI reads and writes — usually a chunk of a word, not a whole word.Large Language Model (LLM)An AI trained on enormous amounts of text to predict the next piece of writing — the technology behind chatbots like ChatGPT and Claude.Retrieval-Augmented Generation (RAG)Letting an AI answer from a specific set of documents by looking them up as it responds — instead of relying only on what it memorized.Fine-tuningContinuing a model's training on your own examples so its behavior changes — baked into the model, not supplied at answer time.Prompt EngineeringThe craft of writing instructions that get the best, most reliable output from an AI model.HallucinationWhen an AI produces something fluent and confident that is simply false — fluency is not the same as accuracy.
View all 40 in Language & LLMsAI Agents
17AI AgentSoftware that pursues a goal by taking its own steps — deciding, acting, and reacting — instead of answering once and stopping.Tool UseLetting a model call real software — a search, a calculator, your database — instead of trying to answer everything from memory.Agent MemoryGiving an AI a way to remember across conversations, since the model itself forgets everything the moment a session ends.Multi-Agent SystemsSeveral AI agents working together on one problem, each with a role — powerful in demos, awkward in production.GuardrailsThe checks around a model that decide what it's allowed to receive, say, and do — the part that stops a demo becoming an incident.Function CallingHow a model asks your code to do something — the mechanism underneath every agent, and it's the model requesting, never executing.
View all 17 in AI AgentsGenerative AI
12Diffusion ModelHow most AI image tools work — starting from random noise and removing it step by step, guided by a prompt, until a picture appears.GAN (Generative Adversarial Network)Two networks trained against each other — one faking, one detecting — until the fakes pass. The technique diffusion largely replaced.Multimodal AIModels that handle more than one kind of input — text and images, sometimes audio and video — in a single shared representation.Text-to-ImageType a description, get a picture that didn't exist — the capability that made AI visible to everyone, and the one with the most unresolved argument underneath it.Text-to-VideoGenerating video from a description — not image generation with more frames, because the hard part is that things must stay themselves.Latent SpaceThe compressed space a model thinks in — where similar things sit close together, and where the famous vector arithmetic works better in demos than in practice.
View all 12 in Generative AIMachine Learning
47Supervised LearningTeaching an AI by showing it labelled examples — inputs paired with the correct answers — so it can predict answers for new inputs.OverfittingWhen a model memorizes its training data instead of learning the general pattern — so it looks great in training but fails on new data.Unsupervised LearningFinding structure in data nobody labelled — useful, underrated, and much harder to know if you got right.ClusteringGrouping things that resemble each other — and the fact that the algorithm always returns groups, whether or not any exist.Feature EngineeringReshaping raw data into things a model can actually use — still where most of the accuracy comes from outside deep learning.Train/Test SplitHolding back data the model never sees, so you can find out whether it learned anything or just memorised.
View all 47 in Machine LearningComputer Vision
16Image ClassificationGetting an AI to look at an image and say what it is — the foundational task of computer vision.Object DetectionFinding where objects are in an image and what they are — drawing a labelled box around each one.Image SegmentationLabelling every pixel rather than drawing a box — what you need when the exact shape matters.OCR (Optical Character Recognition)Turning pictures of text into text — solved for clean documents, still genuinely hard for everything else.Vision TransformerCut an image into patches, treat them as words, run a transformer — which works, and only if you have enough data.Face RecognitionIdentifying a person from their face — technically solved, and the single clearest case of a system that works well on average and fails on specific people.
View all 16 in Computer VisionSafety & Ethics
34AI AlignmentThe problem of making AI systems actually do what people intend — reliably pursuing the goals we want, not just the ones we accidentally specified.Bias & FairnessThe problem of AI systems producing unfair or discriminatory outcomes — usually by absorbing biases present in their training data.ExplainabilityGetting a model to show its working — and the uncomfortable fact that most methods explain the explanation, not the decision.JailbreakingGetting a model to do what it was trained to refuse — and the structural reason it keeps working.Red-teamingAttacking your own system on purpose, before someone else does it for free.Privacy & PIIPersonal data going into AI systems, coming back out of them, and the fact that a trained model is very hard to un-train.
View all 34 in Safety & EthicsFoundations
59Machine LearningGetting computers to learn patterns from data and improve at a task, instead of being explicitly programmed with rules.Deep LearningMachine learning using neural networks with many layers — the approach behind nearly every recent AI breakthrough.Reinforcement LearningLearning by trial and error through rewards — the way you'd train a pet, applied to software.Artificial IntelligenceThe field of making machines do things that seem to require intelligence — a definition that has moved every time the machines succeed.AGI (Artificial General Intelligence)A hypothetical system with broad human-level capability across domains — undefined enough that people can argue about whether it's arrived.Training vs InferenceBuilding the model versus using it — two completely different activities with different costs, hardware, and constraints.
View all 59 in FoundationsTools & Ecosystem
16Vector DatabaseA database built to store embeddings and find the most similar ones fast — the search engine behind meaning-based retrieval.Open-Weight ModelsModels whose weights you can download and run yourself — often called open source, usually not quite.GPUThe chip that made deep learning possible — thousands of small cores doing the same maths at once, which is exactly what neural networks need.Inference APIRenting a model by the request — how nearly everyone actually uses AI, and the dependency that comes with it.Semantic SearchSearching by meaning rather than by words — which finds what keyword search misses, and misses what keyword search finds.Knowledge GraphFacts as a network of entities and relationships — symbolic AI's one commercial survivor, and RAG is rediscovering it.
View all 16 in Tools & EcosystemSpeech & Audio
13Speech RecognitionTurning spoken audio into text — solved for clear speech in quiet rooms, and still genuinely hard for everything real.Text-to-SpeechTurning text into speech that sounds human — where the remaining gap isn't the voice, it's knowing which word to stress.Voice CloningCopying a specific person's voice from a short sample — technically impressive, ethically unresolved, and already being used against people.Speaker DiarizationWorking out who spoke when — the unglamorous half of transcription, and usually the half that's wrong.Music GenerationModels that produce music from a description — good enough for background, and sitting on an unresolved argument about whose work it learned from.SpectrogramTurning sound into a picture so a vision model can look at it — the representation nearly all audio AI runs on, and it throws half the signal away.
View all 13 in Speech & AudioApplied AI
21Recommender SystemThe AI that decides what you see next — probably the most economically significant machine learning on earth, and the least discussed.Time Series ForecastingPredicting what comes next in a sequence over time — where simple methods beat sophisticated ones for forty years, and only recently stopped.Anomaly DetectionFinding the unusual thing — where the base rate makes precision nearly impossible and almost every deployment drowns in false alarms.Named Entity RecognitionPulling the names, dates and places out of text — reported as solved, and reliably disappointing on anything that isn't news.Sentiment AnalysisDeciding whether text is positive or negative — the most deployed NLP task, and the one whose target may not exist.Machine TranslationTranslating between languages automatically — the task that invented modern NLP, where fluency arrived long before reliability.
View all 21 in Applied AI