How this site works

What it is for

This site concisely explains key ideas in AI safety for readers without a technical background. The aim is to help you judge the arguments for yourself, not to persuade you of a particular level of concern.

How each explainer is built

Most explainers have the same parts. “What it is” defines the idea. “Why it matters” explains its role in arguments about AI risk. “The evidence so far” reports what research has actually found.

Standards of evidence

Claims link to their sources, and primary sources (papers, system cards, official documents) are used wherever possible. Each explainer separates what is established from what is contested and what is still open, and it reports evidence against a concern as carefully as evidence for it.

Keeping it current

Research on AI moves quickly, so the evidence sections will date. I will try to consistently update them when important new findings appear, and each page shows when its evidence was last checked.

Corrections and AI assistance

If you spot an error, please get in touch through the About page. I use AI tools to help prepare drafts, but every explainer is thoroughly edited and carefully checked by me before it is published.