The AI Fix ChatGPT Health, Victorian LLMs, and the biggest AI bluffers
25 snips
Jan 13, 2026 In this episode, a marriage is declared invalid due to vows drafted by ChatGPT—an eye-opening legal glitch! The hosts dive into the quirky world of AI bartenders and review Glide's AI hair clippers. Graham shares a story about a journalist who exposed AI bluffers with a fake idiom, while Mark hilariously discusses his New Year's resolution of lying to LLMs. They also explore how fine-tuning models can yield unexpected biases, and what happens when they respond like they're from the Victorian era. It's a blend of laughs and insights!
AI Snips
Chapters
Books
Transcript
Episode notes
ChatGPT Health Offers Personalised Records
- OpenAI launched ChatGPT Health to connect medical records and apps for personalised health conversations.
- OpenAI says this data won’t be used for model training but privacy concerns remain for many listeners.
Stack Overflow Reinvented Around AI
- Stack Overflow question volume plunged after ChatGPT's launch, but the company grew revenue by selling its data and AI tools.
- Stack Overflow monetised by licensing content and selling Stack Internal to thousands of companies.
Lying To AIs As A New-Year Game
- Mark Stockley resolved to lie to AIs and describes making up diplomatic and musical claims to fool chatbots.
- He shares examples like being a deputy ambassador to Finland and inventing a song to probe AI responses.
