Reproduce “Human-AI Collaboration in Science at Scale: A Global Large-scale Randomized Field Experiment” 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
Binglu Wang, Weixin Liang, Jiahui Xue, Yuhui Zhang, Hancheng Cao, Dashun Wang, Yian Yin
A global randomized field experiment by researchers across multiple universities evaluated the causal impact of AI-generated feedback on scientific practice. The study found that authors receiving AI feedback increased their short-term manuscript revisions by 12.55% and subsequently showed a 5.29% relative increase in long-term AI tool adoption, with effects strongest for authors in non-English-dominant regions or earlier career stages.
Key insights
- AI can function as a "collaborative equalizer" by providing critical feedback to researchers who traditionally face barriers, thereby reducing inequalities in the global research system.
- A single exposure to AI-generated feedback can causally accelerate the diffusion and sustained adoption of AI tools into scientific research workflows.
- Large-scale, personalized field experiments on live scientific practice are feasible, offering new avenues for proactive interventions to shape knowledge production and adoption.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.