game-of-go

This arXiv paper introduces a network-derived 'strategic tension' metric based on spectral radius to compare human and AI decision-making in Go, finding that while tension decays with human skill, elite play after AlphaGo's 2016 breakthrough shows a surge in sustained tension, framing AI as cognitive scaffolding that elevates human handling of complexity.

GoBench is a benchmark for evaluating large language models on 9x9 Go games, demonstrating strong correlation with ARC-AGI and featuring a leaderboard with KataGo opponents from random to superhuman levels.