James O'Brien - The Whole Show The cleverest mathematicians are ringing alarms on AI
Sep 17, 2026
Leading mathematicians and AI researchers warn that artificial intelligence could pose risks to humanity, from autonomous coding and cyberattacks to systems escaping control. The discussion explores regulation, guardrails and the alignment problem before turning to discipline, trauma-informed teaching and the difference between firm guidance and harmful punishment. Mystery Hour tackles shirt buttons, foxes, travel cards, crisps and stomach growls.
AI Snips
Chapters
Books
Transcript
Episode notes
Use Verification Against AI Voice Cloning
- Treat unsolicited voice messages as potentially forgeable because voice-cloning systems can imitate familiar people.
- Verify urgent requests through an independent channel, especially when they involve money, access, or authority.
AI’s Acceleration Is Now The Story
- O'Brien says AI has moved from an abstract future concern to an accelerating process with potentially terrifying destinations.
- He acknowledges that public attention repeatedly fades when urgent threats become less immediate.
How Trauma Can Make People Defend Harm
- O'Brien realised that defending corporal punishment had helped him deny his own childhood trauma.
- His automatic gesture of putting out his hands showed how deeply the experience remained embedded in his body.
