Benjamin Todd: AI progress beat superforecasters over 18 months
ben_j_todd · x · 2026-10-26
Benjamin Todd reviews 18 months of AI progress and argues almost everyone underestimated it. Superforecasters and the public were furthest off; AI experts were closer but still too slow; only company leaders were slightly too optimistic. His own 2025 "AGI by 2030?" piece turns out to have been too conservative: the four drivers (pretraining, RL, test-time compute, agent scaffolding) are holding, but his trend-continuation capability numbers for end-2026 were too slow. Useful calibration: assume what CEOs say will happen in six months takes six to 18. He also notes robotics forecasts failing—median 2034 for a humanoid beating a human over 5km, yet humanoids broke the human half-marathon record in April 2026 and ran 100m faster than Bolt in August.
Related event: Ben Todd: AI Wasn't Overhyped—Almost Everyone Underestimated It(5 posts)→
More from AGI Musings
- dunkhippo33 Shares How He Thinks About Raising Kids in the AI Era — kimmonismus · 2026-10-27
- EE-trained AI educator: systems-bottleneck thinking still matters in the AI era — dunkhippo33 · 2026-10-27
- Why Americans pitch better: a lifetime of class presentations, says AI educator — dunkhippo33 · 2026-10-27
- 'Half the sites I try via agents block access' — the anti-bot paradox holding back the agentic economy — XFreeze · 2026-10-27
- Commoditization dilemma: either human labor or GPUs become the commodity — _AustinCalvert_ · 2026-10-27
- Cosmos Institute launches Oxford AI philosophy journal with $10,000 best-paper prize — lawhsw · 2026-10-27