EBR-bench Study: AI Lacks Human-like Continual Learning in Long-horizon Tasks
Jsevillamol · x · 2026-08-29
Research by Epoch AI using the complex long-horizon board game Earthborne Rangers and its benchmark EBR-bench compares the continual learning abilities of humans and AI.
- Human Performance: Humans start low but, through experience, the best players master the game after about 5 playthroughs.
- AI Performance: None of the evaluated AI models reached the level of human mastery, failing to show significant improvement with experience.
- Conclusion: This provides evidence that the problem of Continual Learning in AI is not yet solved, with AI showing significant gaps in human-like learning curves and adaptation to long-horizon tasks.
Related event: Epoch AI study finds AI lacks human-like learning through repeated attempts(2 posts)→
More from Research
- False Collatz Proof via a Lean Bug: Talk Probes Proof Assistants as LLM-Found Bug Targets — marc_lelarge · 2026-10-16
- Goodfire's Nick Cammarata argues neural networks are 'already clean' in new interpretability essay — ctjlewis · 2026-10-16
- NEJM AI: prompt tuning cuts LLM clinical test-order cost ratio to 1.4x across 19 systems — dvir_a · 2026-10-16
- AI Recommends 3.2x Costlier Care; Models Add Tests in 2/3 of Visits Physicians Ordered None — dvir_a · 2026-10-16
- Prompting cuts LLM over-testing costs across 19 systems, but diagnosis agreement drops — dvir_a · 2026-10-16
- 24 AI Systems Recommend $314 Per Visit in Care vs Physicians' $97, a 3.2x Gap — dvir_a · 2026-10-16