"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis cover image

Reward Hacking by Reasoning Models & Loss of Control Scenarios w/ Jeffrey Ladish of Palisade Research, from FLI Podcast

"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis

00:00

Challenges in Training AI for Long Horizon Tasks

This chapter examines the difficulties of training AI for extended tasks that span days or weeks. It discusses methods for acquiring training data, emphasizes the importance of breaking down tasks into sub-steps, and reassures that these challenges can be overcome with sufficient resources and engineering.

Transcript
Play full episode

The AI-powered Podcast Player

Save insights by tapping your headphones, chat with episodes, discover the best highlights - and more!
App store bannerPlay store banner
Get the app