Reproduce “CuriousTogether: Implications of Curiosity and Cultural Grounding in Multi-Agent LLM Interactions” 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
Angana Borah, Zoha Hayat Bhatti, Zhijing Jin, Rada Mihalcea
CuriousTogether studies whether question-guided discussion helps multi-agent groups across creative, reasoning, and social tasks. Three agents exchange questions and may revise over two rounds, with prompts encouraging exploration of assumptions, evidence, alternatives, or perspectives. Curiosity-guided settings improve numerically on seven of eight outcomes; the anti-stereotype score rises 14.90 percentage points versus no-curiosity settings, though only three comparisons survive Holm correction.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.