A study analyzing chess positions found that even when engines like Stockfish 18 deem them equal, human players exhibit consistent outcome skews, favoring either White or Black. These skews were reproducible across different player groups, time periods, and rating bands, and persisted even in highly popular and well-measured positions. While the typical skew is small, the disfavored side tends to spend more thinking time, suggesting that engine evaluations alone do not fully predict human game outcomes. AI
IMPACT Suggests that AI evaluations in complex domains may not fully capture human behavior or outcomes.
RANK_REASON Academic paper detailing a reproducible outcome skew in engine-assessed equal chess positions. [lever_c_demoted from research: ic=1 ai=0.4]
Read on Hugging Face Daily Papers →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
