Gary Marcus’s 2022 AI predictions already look outdated by 2026
dhadfieldmenell · x · 2026-07-24
A 2022 Gary Marcus prediction list looks increasingly wrong in 2026
The thread revisits a 2022 set of predictions by Gary Marcus about what AI would not be able to do by 2029. The poster’s takeaway is that several of those claims already look “solved” or close to it by 2026, even though one or two remain uncertain.
The tone is explicitly not a dunk; it frames Marcus as having made concrete, falsifiable predictions and says reality has moved faster than expected. The attached image shows Marcus’s original list, including limits around movie comprehension, novel reading, cooking, bug-free 10k-line code generation, and converting natural-language math proofs into symbolic form.
More from AGI Musings
- NeurIPS 2026 workshop will spotlight failure modes of AI in biology — anshulkundaje · 2026-07-24
- Metascience could reshape the scientific core, not replace it — anshulkundaje · 2026-07-24
- Frontier models still depend on stable university research and global talent — anshulkundaje · 2026-07-24
- AI industry pulls professors out of academia as research funding shrinks — ruthstarkman · 2026-07-24
- Token efficiency comes from search and validation, not just implementation — yunta_tsai · 2026-07-24
- Open-weight frontier models could become dangerous if they can be jailbroken and used anonymously — Afinetheorem · 2026-07-24