Episode Details

Move over, stochastic parrots—the era of the deliberative machine has arrived. We dive deep into the shift from "fast and wrong" to "slow and right," exploring how inference-time scaling and System 2 thinking are transforming AI from a guessing machine into a reasoning powerhouse. From the secrets of hidden thought tokens and GRPO to the risks of alignment faking, we dissect whether this cognitive leap is a productivity miracle or a transparency nightmare.