Why AI forgets you between conversations, and what to do about it2 September 2026
Explainers
Two series. How the Models Work explains the technology in plain language. The Measured View works through the evidence, one study at a time.
How the Models Work
Explainers of how AI systems actually work, written for people who make decisions about them. Each piece takes one concept and explains it plainly, with no jargon left undefined.
Why AI makes things up about your business, and how RAG fixes it21 July 2026
Why prompt injection is the number one AI risk, and what actually helps20 July 2026
Why AI hallucinates, and how to catch it16 July 2026
Why a bigger context window is not simply better15 July 2026
Why the newest AI models are slower on purpose13 July 2026
The Measured View
Analysis of published AI research, one paper at a time. What was measured, what the numbers show, what the study cannot support, and what it means for people running organisations.
The black box has seams: what Anthropic's circuit tracing mapped21 August 2026
Showing its work: what Anthropic's faithfulness study measured3 August 2026
The complexity cliff: what Apple's puzzle study measured1 August 2026
The expert line: what OpenAI's GDPval benchmark measured30 July 2026
The AI coding slowdown: what METR's controlled trial measured29 July 2026