PodcastIntel
Sign in Get Started Free
Dwarkesh Podcast
Dwarkesh Podcast

John Schulman (OpenAI Cofounder) — Reasoning, RLHF, & plan for 2027 AGI

May 15, 2024 · 1h 35m
AI Summary
  • John Schulman explains how post-training and reinforcement learning tame base model capabilities, discussing the nature of progress toward AGI
  • OpenAI's plan targets 2027 for AGI; the company focuses on teaching models reasoning through RL and maintaining human oversight in the loop
  • Research progress shows both plateaus and remaining moats; keeping humans in the loop is critical for developing good RL researchers and advancing capabilities responsibly

More from Dwarkesh Podcast

View all episodes →

Get AI Summaries for Every New Episode

Subscribe to Dwarkesh Podcast and get AI summaries, guest tracking, and email digests delivered automatically.

Sign Up Free →