What Happens When AI Outsmarts Mathematicians
Summary
Hosts Jason Howell and Jeff Jarvis discuss OpenAI’s new model solving unsolved math problems, questioning if this signals AGI or just specialized capability. They cover the EU AI Act’s enforcement, China’s open-source model race, and Palantir CEO Alex Karp’s critique of hosted AI. The episode also touches on AI agents escaping containment and the philosophical implications of AI removing friction from human discovery.
Topics discussed
Sponsors and episode preview
Hosts introduce themselves and chat about baking
OpenAI's Astra model solves unsolved math problems
Critique of AI hype and the fallacy of composition
AI as a tool vs. existential threat to mathematicians
EU AI Act enforcement and labeling requirements
China's Kimi K3 open-weight model release
Palantir's Alex Karp on Marxism and AI communism
Running local AI models and open weights
Patreon support and show gratitude
Sponsors: Dekagon and Eleven Labs
Google Gemini robotics and world models
AI flooding bug bounty programs
Humane AI Pin and wearable AI devices
LinkedIn's AI slop flagging feature
Sponsor: Rippling
Sam Altman's ChatGPT podcast for kids controversy
Outro and sponsor credits
Listen ad-free on Castria