AI commentator doubles down: model progress 'all true', citing Terence Tao's o1 graduate-student benchmark
RichmanRonald · x · 2026-09-23
RichmanRonald endorses @AndrewCurran's thread arguing AI progress is systematically underestimated. The author stresses he takes no money from labs and posts only to warn about what's coming.
The thread's key evidence: Terence Tao, roughly two years ago on OpenAI's o1, described the experience as like advising a 'mediocre, but not completely incompetent, graduate student' — a clear step up from prior models. Every recent advance, the author argues, is another piece of the same pattern.
More from AGI Musings
- Major model release gap shrinks from 73 days in 2023 to 18 days in 2026 — MickeySteamboat · 2026-09-23
- Schmidhuber Riffs on Ezra Klein's Call to Ban AI Recursive Self-Improvement — SchmidhuberAI · 2026-09-23
- Even an exponential AI takeoff won't break humanity, argues one X poster — zeeshanp_ · 2026-09-23
- Schmidhuber: I Published the First Concrete Recursive Self-Improvement Algorithms Back in 1987 — SchmidhuberAI · 2026-09-23
- GPT-6 Astra Now Automates 20.8% of Remote Work, Up From 2.5% Last Year — Puzzleheaded-King584 · 2026-09-23
- Few People Have Internalized That Longevity Escape Velocity Is Near — LesaunH · 2026-09-23