GameLab’s Human Gameplay Data is one of the largest continuously growing datasets of real human decision-making in interactive environments. Every session captures synchronized action logs, visual game states, and outcomes, all tied together through a unified structure that allows precise reconstruction of how decisions unfold over time. This creates a uniquely clean and scalable dataset for studying behavior across perfect information, imperfect information, stochastic systems, and long-horizon planning tasks.

What makes this dataset valuable is not just its scale, but its fidelity and structure. Each data point reflects real human reasoning under real constraints, uncertainty, risk, and strategy, making it highly applicable beyond games. The dataset serves as a foundation for training, evaluation, and benchmarking, enabling researchers to model cognitive processes, compare AI performance against human baselines, and explore generalizable decision-making capabilities.