Papers
arxiv:2608.14667

Position: AI Agents in Scientific Teams Should Be Studied as Human-Agent Systems

Published on Aug 2
· Submitted by
Leo Y
on Aug 18
Authors:
,
,
,
,
,
,
,
,
,
,
,
,
,

Abstract

Scientific collaboration with AI agents requires studying human-agent pairs to avoid risks like reduced inquiry diversity and to foster synergistic discovery.

Large language model-based agents are increasingly deployed as collaborators in scientific discovery yet most current work focuses on the autonomous capabilities of "AI Scientists". We argue that this overlooks the social aspects of scientific teamwork, and that studying AI Scientists as human-agent systems (HAS)--where the unit of analysis is the human-agent pair--is both underexplored and undervalued. We establish these points through literature and empirical analysis, and highlight recent incidences and studies which show that deploying agents in science without accounting for human-agent dynamics introduces near-term risks, including reduced diversity of scientific inquiry. Through analysis of real-world case studies, we show that scientists and agents can augment each other's capabilities. We call for new research that adopts the HAS lens to develop mathematical frameworks for understanding and fostering human-AI synergy in scientific discovery.

Community

Paper submitter

This position paper argues that AI agents used in scientific discovery should be studied as human-agent systems, not just evaluated on their standalone capabilities. The authors review current "AI Scientist" systems and find most only let humans set goals and check final outputs, missing the back-and-forth collaboration real science teams rely on.

They point to concrete risks of this gap: hallucinated citations showing up in accepted NeurIPS and ACL papers, a documented drop in topic diversity when AI assists research, and signs of de-skilling among junior scientists who lean too heavily on AI tools. Two case studies (writing up a theorem with Gemini 3 Pro, and modeling thermonuclear burn with GPT-5) show how humans and agents can genuinely complement each other when experts stay actively involved.

The paper ends with a simple framework for measuring collaboration benefit versus coordination cost, and calls for more research into human-AI synergy specifically for scientific discovery.

Sign up or log in to comment

Models citing this paper 0

No model linking this paper

Cite arxiv.org/abs/2608.14667 in a model README.md to link it from this page.

Datasets citing this paper 0

No dataset linking this paper

Cite arxiv.org/abs/2608.14667 in a dataset README.md to link it from this page.

Spaces citing this paper 0

No Space linking this paper

Cite arxiv.org/abs/2608.14667 in a Space README.md to link it from this page.

Collections including this paper 0

No Collection including this paper

Add this paper to a collection to link it from this page.