When AI Overthinks: Lessons from the Apple Illusion of Thinking Paper
This week, Rob May, Calvin Cooper, and Neurometric CTO Byron Galbraith unpack Apple’s Illusion of Thinking paper and what it tells us about the limits of reasoning models.
We cover:
Why the paper split the AI community
What hidden “thinking traces” reveal about performance
Why models loop, stall, or abandon tasks
An open-source tool we built to replicate the study
Listen here: https://inferencetimetactics.podbean.com/?utm_campaign=as-npc105112516