PodcastIntel
Sign in Get Started Free
Future of Life Institute

Why AI Evaluations Are Broken and How to Fix Them (with David Manheim)

Jul 17, 2026 · 1:19:21
AI Summary
  • AI evaluations are unreliable due to unclear reporting and 'training to the test'.
  • Fixes include real-world tests, transparency, and human oversight.
  • Even normal AI progress could be disruptive and requires careful evaluation.

Guests on This Episode

More from Future of Life Institute

Aug 13, 2026 · 0:25:55
View all episodes →

Get AI Summaries for Every New Episode

Subscribe to Future of Life Institute and get AI summaries, guest tracking, and email digests delivered automatically.

Sign Up Free →