Did Demis Hassabis bet wrong in the AI race? Reddit post dissects DeepMind's slide
Recent_Fox4339 · reddit · 2026-09-24
A Reddit long-post questions whether Demis Hassabis made the wrong bets, as OpenAI's Navier–Stokes result and Anthropic's biology breakthrough land instead of DeepMind's.
Possible explanations discussed:
- Over-betting on world models and spreading across too many areas while underestimating LLM scaling
- Friction from the GDM merger: London research culture vs Mountain View engineering, plus rumored tension with Brin (who crashed a Demis interview)
- Attention divided by AlphaFold and Isomorphic Labs
- Reported lack of appetite for the commercial race (2017 spin-out attempt 'Project Mario')
- The breakup of the AlphaFold team: John Jumper moved to coding work, then left for Anthropic
Author guesses Demis leaves Google within a year or two and GDM gets fully absorbed, while admitting he may be judging the race too early.
More from AGI Musings
- Mathematicians face an identity crisis as AI solves problems no one can verify — ValerioCapraro · 2026-11-11
- Why inference speed will be AI products' biggest competitive edge in 1–2 years — itsOmSarraf_ · 2026-11-11
- Ben Todd's alt log-log plot shows task horizons hitting a singularity in under 12 months — ben_j_todd · 2026-11-11
- AI risk community splits over whether the intelligence explosion arrives in 2027 — sebkrier · 2026-11-11
- The Zvi on unstrategic AI predictions: why overhyping medical AI undermines AI risk credibility — TheZvi · 2026-11-11
- Comic essay on intellectual humility: meaning may exceed human cognition — YogeshMalik · 2026-11-11