Reproduce “A Foundation Model to Measure and Predict Human Behaviour” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Haoran Ye, Tianrui Huo, Hangzhen Li, Xin Zhang, Guojie Song
Chiron trains one language model to measure psychological constructs from personal records, predict withheld answers, and simulate populations with synthetic respondents calibrated to human statistics. Across held-out studies, its mean absolute error for standardized effect sizes was 0.224, versus 1.254–3.895 for frontier models; this average covers 25 effects from eight of 15 studies. It outperformed baselines on four of five population metrics, but human confirmation remains necessary.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.