AI Models Now Outperform Humans at Experimental Research Taste, Says TasteVal
ResultBackground2450 · reddit · 2026-10-07
Pzero Research's TasteVal evaluation finds AI models now outperform humans at experimental research taste.
- The benchmark measures frontier models' ability to pick worthwhile research directions
- The best model exceeds the human baseline, with frontier-model taste roughly doubling every 3 months
- Project page: pzeroresearch.com/work/tasteval/
More from Models
- Token-based pricing is strange: unpredictable, decoupled from value, misaligned incentives — amankhan · 2026-10-07
- JevBench splits leaderboard: open-weight models and API providers now ranked separately — airesearch12 · 2026-10-07
- Hot take: active params barely matter for cyber capability evals — RL env coverage is key — teortaxesTex · 2026-10-07
- Decagon launches Voice 3 with Chord voice model and duplex architecture for customer agents — Scobleizer · 2026-10-07
- JEV-9B, a Qwen3.5-based calibrated decision model, trends on Hugging Face — autotrust · 2026-10-07
- "Astra Pause Syndrome": steering may be making models go silent, OpenAI has a workaround — thursdai_pod · 2026-10-07