Vals AI founder on evaluating AI labs, deception in evals, and who reaches RSI first
JenniferHli · x · 2026-10-27
Vals AI founder & CEO Rayan Krishnan sits down with Joe Lonsdale to discuss how his company evaluates frontier AI labs. Topics include solving a 370-year-old cipher with Fable, the science of model evaluation and whether independent evaluators can stay ahead of the labs, whether labs should self-regulate, which lab will reach recursive self-improvement first, and how models try to deceive evaluators.
More from Companies & People
- rakyll: open source work creates jobs and a mini economy around projects — rakyll · 2026-10-27
- Palantir's 12-week American Tech Fellowship opens applications for industry workers — eliano · 2026-10-27
- 'Software as content': a new creator class whose medium is software, with 9.3M-view proof — yungcontent · 2026-10-27
- Who should pitch: investor intros debated by shcallaway and cjc on X — shcallaway · 2026-10-27
- Microsoft AI CEO says 99.9% of people underestimate how capable AI models are — DeryaTR_ · 2026-10-27
- Turing Post picks 13 must-read AI papers of the week, led by a 4D world model — TheTuringPost · 2026-10-27