Gary Marcus reviews 2025 AI predictions: 17/18 correct on scaling hitting a wall
GaryMarcus · x · 2026-11-04
Defending against critic Sabine Hossenfelder (@skdh), Gary Marcus links his 2025 predictions review, claiming 17/18 high-confidence calls were correct: still GPT-4-level models, no GPT-5, price wars, thin moats, unsolved hallucinations, limited enterprise adoption, most model companies losing money. His 2022-era claims that pure LLM scaling would run out and AI was 'hitting a wall' have, he argues, been confirmed by industry insiders in recent months.
Related event: Gary Marcus Claims 17/18 Predictions Right in Feud with Sabine(3 posts)→
More from AGI Musings
- State of AI author Nathan Benaich on recursive self-improvement and why AI-native firms grow 3x faster — arena · 2026-11-05
- Luiza Jarovsky: AI labs are co-responsible for rising AI cyberattacks and should be legally required to help defend society — LuizaJarovsky · 2026-11-05
- CellType claims 27B biology model beats all 12 frontier models at predicting cell behavior — alifcoder · 2026-11-05
- AI circle debates Dario's 'Machines of Loving Grace' and EA-shaped Claude — teortaxesTex · 2026-11-05
- Inside a16z's agent podcast: trust is the top fear among 147 tracked AI assistants — a16z · 2026-11-05
- Who gets to be expensive? Agents get subsidized while companion users get rationed — serialchilla91 · 2026-11-05