Reproduce “Exploring The Impact Of Proactive Generative AI Agent Roles In Time-Sensitive Collaborative Problem-Solving Tasks” 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
Anirban Mukhopadhyay, Kevin Salubre, Hifza Javed, Shashank Mehrotra, Kumar Akash
This study investigated how proactive generative AI agents, acting as either a peer or a facilitator, influence human collaboration in time-sensitive problem-solving tasks. Researchers found that a facilitator AI agent led to the highest group performance, while a peer AI agent resulted in increased workload and lower performance, highlighting a divergence between perceived AI utility and objective outcomes.
Key insights
- A significant disconnect was observed between perceived AI helpfulness and objective performance, where the subtle facilitator agent correlated with highest group performance despite being less salient, while the seemingly influential peer agent correlated with lowest performance and higher workload.
- Proactive AI roles present distinct challenges in time-sensitive settings: facilitators risked redundancy and distraction, while peers, though offering useful hints, often disrupted collaborative flow, increased cognitive burden, and fragmented communication.
- Human-AI collaboration is dynamic, with varied trajectories of agent use ranging from initial reliance leading to reflection, to early enthusiasm turning into dependence and disillusionment, or quickly leading to frustration and disengagement.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.