This week, the verdict on the expensive AI I switched off: the free engine wins, though not for the reason I hoped. A machine of mine that reported success every night for a week while doing nothing at all, and what it taught me about trusting a green tick. One idea from the book on the confident fool that can never say "I don't know." Plus the week's AI news, checked before it's aired.
Chapters:
00:00 Cold open
00:57 Intro
01:50 Build Log, the verdict + the silent robot
04:54 The Bookshelf, The Confident Fool
07:29 AI News (verified)
11:08 Outro
AI-voice disclosure: Narrated with an AI-generated clone of Peter's own voice. Written and edited by a human. The AI-news segment is fact-checked against primary sources before airing.
Featured app, Utter List: https://apps.apple.com/gb/app/utter-list/id6761653207
TabletMinder, all 50,000+ NHS medicines built in: https://apps.apple.com/gb/app/tabletminder/id6757861874
Book, Understanding AI: The Word Calculator and How to Use It Well (Peter Hunter): https://books2read.com/u/31zr9r
AI-news sources:
Anthropic Developer Platform, expiring API keys (2026-07-08) + Admin API beta (2026-07-14): https://platform.claude.com/docs/en/release-notes/api
Claude Code changelog, auto-background long MCP calls (2026-07-17) + Auto mode on Bedrock/Vertex/Foundry (2026-07-11): https://code.claude.com/docs/en/changelog
(Held, not aired) OpenAI GPT-5.6 "Sol", primary source still could not be verified (HTTP 403): https://openai.com/index/previewing-gpt-5-6-sol/
Narrated with an AI-generated clone of Peter's own voice. Written and edited by a human. The AI-news segment is fact-checked against multiple independent sources before airing.
