Three frontier models graded 50 tech leaders' public speech — LeCun ranks #1
tonygwu · x · 2026-10-01
The author used three frontier models (Fable, Astra, Gemini) as independent judges to rank 50 tech leaders based solely on verbatim transcripts — 653 long-form podcasts, interviews and keynotes totaling 8.8M words — scoring Insight (45%), technical/industry depth (35%) and clarity (20%).
Yann LeCun came out #1, followed by Jeff Bezos and Tim Sweeney; Marc Benioff ranked last at #50.
The methodology is surprisingly rigorous: each judge grades transcripts independently against a 15-criterion rubric, with inter-judge agreement of 0.854r; gaps smaller than 4.3 points are treated as statistical ties, and every score carries the judge's written reasoning and cited evidence. Code, rubric and tests are open-sourced on GitHub.
Related event: LLM blind review of 8.8M interview words ranks LeCun top for insight(3 posts)→
More from Fun
- 8-Year-Old Teaches Engineering Students to Build Games with ChatGPT, Claude and Codex — cneuralnetwork · 2026-10-01
- NYT sues OpenAI — while its own headline turns out to be AI-generated for SEO — TheTuringPost · 2026-10-01
- Easter egg: liking any Grok Bot post on X triggers a surprise — soleio · 2026-10-01
- DevDay demo: Codex builds Minecraft for 30-year-old Game Boy hardware via ModRetro plugin — pvncher · 2026-10-01
- FuhuihuaBench meme benchmark: every model scores Pass@16 = 0 — bowenc0221 · 2026-10-01
- AI Kept Saying Ice It; ER Doctor Said Heat, Swelling Gone Next Day — DesireeCachette · 2026-10-01