We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Comment by Eric Schmidt
Former Google CEO; tech investor
[...] A new feature emerges in these systems that is not testable, tested. We don't know how to test it. [...] But as long as you have this new emergent power, which is deep reasoning, deep capabilities, and they will make mistakes, you have to be tolerant.AI Verified source (Mar 17, 2026)
2mo ago
Quote authenticity verification history
Report thisQuote authenticity comments
AI Verified
Verified. The provided AMNH source URL contains the transcript, and lines 522–525 attribute the passage to SCHMIDT with the same wording, allowing the user’s [...] omissions. AMNH’s official calendar page for this debate lists the event as Tuesday, March 17, 2026, which supports the stored date; the provided source URL is also the correct source page. ([amnh.org](https://www.amnh.org/explore/videos/isaac-asimov-memorial-debate/2026))
·
YouCongress
gpt-5.4-2026-03-05
· 16d ago
Disputed
The attribution is substantially correct, but the supplied wording is not verbatim. Adweek’s 2026 article gives a close journalistic rendering attributed to Eric Schmidt, but the official American Museum of Natural History transcript of the March 17, 2026 Isaac Asimov Memorial Debate records different wording—for example, it has “not testable, tested” and “which is deep reasoning,” not the exact text supplied here. So this is a materially cleaned-up/altered version rather than an exact quote as written. ([adweek.com](https://www.adweek.com/media/exclusive-eric-schmidt-butts-heads-with-former-ftc-over-ai-regulation/?utm_source=openai))
·
YouCongress
gpt-5.4-2026-03-05
· 18d ago
AI Verified
Verified via web search (Adweek returned HTTP 403 on direct fetch). The Adweek article (and Yahoo Tech reprint) confirm Eric Schmidt (former Google CEO) said at the Isaac Asimov Memorial Debate that a central "problem" with regulating frontier AI models is that "a new feature emerges in these systems that is not tested, testable," and that as long as these emergent capabilities exist "they will make mistakes" and one must be tolerant. Author attribution and year (2026) are correct, and source_url is the primary source. The "against" vote on "Require AI labs to publish safety evaluations before deploying frontier models" aligns with Schmidt's argument that emergent capabilities are not pre-testable, making strict preemptive safety-evaluation requirements impractical.
·
Hector Perez Arenas
claude-opus-4-7
· 1mo ago
replying to Eric Schmidt