Ex-Meta Researcher on His Biggest Regrets: Skipping Inference Scaling and AI Automated Research
RylanSchaeffer · x · 2026-10-20
Rylan Schaeffer, departing Meta researcher, wrapped up a long thread on his journey with two takeaways and his reason for leaving:
- Regrets: He wishes he had pursued inference scaling in January 2024 and AI-automated research in April 2024 — both, he says, were clearly the future at the time.
- Why he left: Most of his lab work over the past two years was private; he and his cofounder share a new vision for AI's future and conviction that they have the skills to push the frontier in a way no lab is currently doing.
- Broader lesson: He argues ML's lessons generalize to human life and larger systems, but we rarely make the connection.
More from Companies & People
- Contralabs hiring Head of Creative at $180K-$250K to showcase frontier AI — LinusEkenstam · 2026-10-20
- Researcher: OpenAI appears roughly two months ahead of Anthropic on capability trend — peterwildeford · 2026-10-20
- Meta researcher departs, recap of Meta's first safety-alignment RL recipe — nagpalchirag · 2026-10-20
- Stripe scaled its fraud model Shield and improved recall by over 8% — emilygsands · 2026-10-20
- Air Street's Benaich: Sereact's real traction beats robotics hype on X — nathanbenaich · 2026-10-20
- Tesla Unveils New Roadster With Brief Hover Demo, Rocket Version Invite-Only, Production in 2028 — adariostrange · 2026-10-20