The ideas behind the stories
Guides to how AI works, how it fails, and what people argue about. Written for a technical reader who is new to the field.
Read them in order, or start with the one you need. Each guide links back to our stories, and the glossary explains the terms you meet in them.
How a language model works
A look inside a language model: what it is as a file, how it splits text into tokens, turns them into numbers, handles long inputs, and uses only some parameters per word.
Model weights, Tokens, Embeddings, The context window, Mixture of experts
Read the guide →How AI models are trained
Follow a model from raw text to helpful assistant: instruction tuning, fine-tuning for specific jobs, learning from scores and human feedback, and copying one model's skills into another.
Pretraining and post-training, Fine-tuning, Reinforcement learning, Distillation
Read the guide →Compute: the chips, clouds and data centers behind AI
This guide explains why AI needs so much specialized hardware, how training and running models cost money, and where the chips, data centers and rented cloud capacity come from.
What compute is and why it scales, GPUs and accelerators, Data centers and power, Training versus inference, Neoclouds
Read the guide →Frontier models, open models, and how they are compared
This guide explains what makes a model frontier, what changes when a model's weights are released openly, and how much you can rely on benchmark scores to compare models.
What counts as a frontier model, Open-weight models, Benchmarks and their limits
Read the guide →Getting answers from a model: prompts, reasoning and routing
This guide explains how models receive standing instructions, why writing out steps can improve answers, what extra thinking time buys you, and how products pick which model responds.
The system prompt, Chain-of-thought, Reasoning effort, Model routing
Read the guide →AI agents: models that take actions
Learn how a language model becomes an agent through tools and surrounding software, why long tasks are hard, and what happens when an agent acts beyond its instructions.
From automation to agents, Tool use and function calling, The agent harness, Computer use, Long-horizon tasks, Rogue agents
Read the guide →
AI safety: a map of the field
This guide maps the main kinds of AI risk, open questions about AI systems themselves, and the research and policy work that addresses them, with links to deeper guides.
What AI safety covers, Misuse and dual-use capabilities, A different kind of question
Read the guide →When AI doesn't do what we meant
This guide explains what it means for an AI model to do what its developers intend, and the ways models fall short, from gaming rewards to hiding their abilities.
Alignment, Reward hacking, Sycophancy, Emergent misalignment, Situational awareness, Scheming and alignment faking, Sandbagging
Read the guide →Checking AI: testing, audits and looking inside models
How researchers test, red-team, audit and look inside AI models, read their written reasoning, and try to prove their behavior, plus what each method can and cannot tell us.
Evaluations, Red teaming, Audits, Interpretability, Chain-of-thought monitoring and why it could be lost, Formal verification
Read the guide →AI security: attacks and defenses
This guide explains how deployed AI models are guarded, how jailbreaks, prompt injection and sandbox escapes get around those guards, and how security flaws and user data are handled.
Guardrails, Jailbreaking, Prompt injection, Sandbox escape, Vulnerability disclosure, Data retention
Read the guide →
Superintelligence and long-term risk
This guide explains what superintelligent AI would be, when and how quickly it might arrive, and the main arguments about whether people could lose control of it.
Superintelligence, AI timelines, Recursive self-improvement, Loss of control, How likely is catastrophe? Existential risk, "P(doom)" estimates, and the main positions
Read the guide →Governing AI: liability, regulation and geopolitics
This guide explains who is responsible when AI causes harm, how the EU, US and China regulate AI, why data centers face local pushback, and how governments control AI chips and compute.
Liability, The EU AI Act, The United States, China's AI rules, Content provenance and watermarking, Data center bans and pauses, Regulatory capture, AI chip export controls, Sovereign AI
Read the guide →Pausing or pacing AI development
This guide explains what pausing or pacing frontier AI development would involve, the main arguments for and against it, and the proposals and groups pushing for it.
A coordinated pause or pacing, The case for a pause, The case against, including the competition-with-China argument, Concrete proposals and who backs them
Read the guide →