When AI gets Personal: Employee emotional responses to anthropomorphic AI agents in a virtual workspace

Published in: Computers in Human Behavior: Artificial Humans — August 2025

Written by

Anand P.A. van Zelderen, Sinuo Wu, Gergely Koszo and Jochen I. Menges

Summary

What we found: We found that employees interacting with anthropomorphic AI agents in virtual workspaces experienced a wide range of emotions, including connection, amusement, and frustration, as well as new emotional responses such as relational assurance and perceived worthlessness. We also discovered that the visual embodiment of AI agents strongly shapes user expectations, where a mismatch between human-like appearance and actual behavior can heighten disappointment.

Why it matters: These findings highlight that AI agents do more than streamline tasks—they actively shape the emotional climate of workplaces. Understanding these dynamics is crucial for organizations aiming to maintain engagement and avoid unintended negative effects on employee well-being.

What next: Organizations and leaders should adopt human-centered design principles when introducing AI agents, ensuring that visual design and functionality align with realistic expectations. They should be mindful that as AI agents become more advanced, the emotional impact they have on workers becomes more pronounced and could erode authentic human experiences.