Reproduce “Think Once, Reflect Twice: How Feedback Shapes Human-AI Collaboration” 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
Joshua Holstein, Philipp Spitzer, Sebastian Schäfer, Maximilian Förster, Gerhard Satzger
The authors test whether feedback on past disagreements with AI advice changes reflection and decisions in loan approvals. After two rounds, participants reviewed successful overrides, unsuccessful overrides, both, or no feedback, then decided again. Balanced feedback was associated with higher self-reported learning than control, without a detected drop in reported deliberation. The brief study did not establish improved performance.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.