Should Models Be Trained on Certain Games? Models May Be 'Too Smart' to Play
scaling01 · x · 2026-09-06
The author questions whether certain games should be excluded from training data, and observes that models may be 'too smart' — of course they know it's just a game — suggesting their metacognition could shape task behavior. An open observation on model behavior and training data choices.
More from Models
- 'SVG, Web Dev, Office Work All Solved' — Call for New AI Benchmarks — BLUECOW009 · 2026-09-06
- Early Gemini 3.8 Flash impressions: pleasant chat and fast, but still weak at coding — zacharynado · 2026-09-06
- "Astra Is a Disaster": User Says New Model Loses Focus, Questions Benchmark-Driven Progress — bambambam7 · 2026-09-06
- GPT-6 Astra hands-on: compute is the bottleneck and CoT observability is eroding — thedealdirector · 2026-09-06
- Yang Zhilin left a US career to build Moonshot AI, now shaking global markets — pstAsiatech · 2026-09-06
- Uncensored models are better writers — safety guardrails make output bland — curious_vii · 2026-09-06