01 / Artificial intelligence
Patterns.
Predictions.
Possibilities.
AI can recognize patterns and generate useful ideas. Understanding how it learns helps us ask better questions about when to trust it. Take the journey: how it works, where it came from, where it is right now, and where it's going.
- Stage
- 3 of 6 · AI agents
- Top of the smarts index
- Anthropic · Claude Opus 5.5
- Top of the people's vote
- Google · Gemini 4 Argon
- Newest frontier model
- Gemini 4 Argon · Sep 30
Open the full tracker →
New here? Start with this
AI in 60 seconds
Five ideas that explain most of what you hear about AI. Each one links to a page that goes further.
It learns from examples. Nobody writes a rule for every spam email. A program studies thousands of emails marked “spam” or “fine” and finds the patterns itself. That is machine learning. Some AI still runs on hand-written rules.
Chatbots write one word at a time. ChatGPT, Claude and Gemini run on large language models, trained on huge amounts of text. Each one picks a likely next word, then the next. That is why an answer can sound sure and still be wrong.
Bigger made it better. Training on far more text, with huge numbers of computer chips, turned rough autocomplete into tools that can translate, summarize and write code.
Now it can take actions. An AI agent works in a loop: plan a step, use a tool such as web search, check the result, repeat. Some 2026 agents keep going for hours. Small mistakes can pile up, so people still check the work.
Next, the physical world. Chatbots mostly know words. Researchers are training world models and robots on video and sensor data so machines can learn how real objects move.
AI foundations
New to AI, or want
the whole picture?
Three ways in. The foundations route explains what different AI methods do, how data becomes numbers, how training differs from using a model, and how to tell whether a system works.
Start from zero
Is every AI a chatbot?
Begin with a map of AI: rules, search, machine learning and deep learning, and how real products mix them. Then follow nine short stops.
Start the foundations →Find a topic or word
All AI topics, one map
Every explainer, lab and game on ARQIV, grouped by what it helps you understand, plus learning paths for different readers.
Browse all topics →Look up a word
AI words in plain language
Short, checked definitions for the words you keep meeting, from tensor and training data to hallucination and RAG.
Open the vocabulary →Go deeper: Evaluating AI · Classical models · Finding groups · Retrieval and RAG · reinforcement learning, embeddings, agents and more
The AI journey
Five stops from
"how?" to "what next?"
Go in order or jump to what you're curious about. Each stop has interactive diagrams that explain one big idea.
How AI works
Neurons, networks and how a machine learns from examples. Try a neuron yourself.
Explore → 02The engineTransformers
Tokens, attention and next-word prediction: the idea behind ChatGPT, Claude and Gemini.
Explore → 031943 → 2026History
Big dreams, two AI winters, and how data and chips made the comeback.
Explore → 04Right nowYou are here
Who's leading the race, and how close are we to human-level AI?
See the tracker → 05The futureWhat's next
World models, robots and the big questions your generation will decide.
Explore →Using AI
Talk to it. Create with it. Connect it.
Now that you know how AI works, learn to use it well: how to write prompts that get great results, how AI makes pictures, video and music, and how it plugs into the rest of the internet.
AI in the world
See AI at work in real science
The same idea, finding patterns in data, is changing every field. Each page has real examples, a hands-on simulator, careers and a classroom activity.
AI in space
How does AI find a planet nobody can see?
Explore →AI in weather & climate
How does AI forecast tomorrow’s weather?
Explore →AI in health
Can AI help a doctor spot what the eye misses?
Explore →AI in engineering & robots
How do a million robots avoid bumping into each other?
Explore →AI in farming & food
How can a camera tell a weed from a crop at 15 mph?
Explore →Put curiosity to work
Read it. Try it. Question it.
Explore new AI research, or take a paper-first investigation into your classroom.