How Employees Appraise and Engage with AI Agents: A Social Exchange Perspective on Human–AI Collaboration

AI agents are increasingly integrated into the workplace as collaboration partners due to their autonomous, human-like capabilities. Such human–AI collaboration (HAIC) can yield beneficial outcomes but relies on employee engagement and willingness to interact. To explain influencing factors, this study proposes the AI Collaboration Acceptance Model (AICAM). It utilizes a framework of Cognitive Appraisal and Social Exchange Theory to conceptualize collaboration relationship value as a key predictor of intention to collaborate. This evaluation, in turn, is driven by secondary appraisals of benefits and costs and primary appraisals of AI agent design cues. The model was validated in an experiment (N = 122) simulating a workplace task with an AI agent. We found that confidence-based trust and perceived similarity with the AI agent strongly drove the evaluation of the collaboration and the intent to maintain it. The results highlight the socio-relational nature of HAIC and offer avenues for practical recommendations.