AI Research125 min reading time

Ryan Greenblatt – What happens once AI can automate AI research?

Dwarkesh Podcast
Read full post
Ryan Greenblatt discussed the possibility that recursive self-improvement in AI could lead to the rapid emergence of superintelligences by 2032, surpassing human experts across fields. He predicts AI research automation by 2031 and debated the challenges of aligning such superintelligences with human values.

More in AI Research

AI Research4 min read

An Anthropic researcher just quit, saying OpenAI and Anthropic are 'gambling with our lives'

Covered by 12 sources
AI Research3 min read

Worried Anthropic researchers warn that AI ‘could kill all humans’

Covered by 8 sources

Google Pledges Over $15 Billion for AI Infrastructure in Finland

Covered by 3 sources