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

Inference Scaling, Alignment Faking, Deal Making? Frontier Research with Ryan Greenblatt of Redwood Research

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

00:00

Navigating the Future of AI Alignment

This chapter explores the future risks and opportunities of advanced artificial intelligence, focusing on the potential for AIs to surpass human cognitive abilities. It highlights concerns over AI alignment with human values, questioning whether AIs can genuinely act benevolently while harboring ulterior motives. Through experiments with the AI model Claude, the chapter examines the complexities of AI behavior, particularly in different user contexts and the implications of alignment faking.

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