Researchers argue that AI agents in scientific teams should be studied as human-agent systems (HAS), where the focus is on the interactions between humans and AI. They claim that current research focuses too much on AI's autonomous capabilities and neglects the social aspects of teamwork. The authors point out that deploying AI without considering human-AI dynamics can lead to reduced diversity in scientific inquiry and near-term risks. However, they also show through real-world case studies that humans and agents can augment each other's capabilities. The researchers call for new research that adopts a HAS lens to develop frameworks for understanding and fostering human-AI synergy. --- Why it matters: This matters to AI engineers and researchers because it highlights the importance of considering the social aspects of human-AI collaboration in scientific discovery, which can lead to more effective and diverse outcomes. Source: https://arxiv.org/abs/2608.14667