The Breakdown

• Researchers studied AI manager clone agents that learn a supervisor's role and assist in managing workplace tasks.
• AI may shift — but not replace — management, handling routine tasks while humans focus on interpersonal responsibilities.
• Study participants saw some potential benefits but worried about impacts on trust, relationships and power dynamics.

 ***

Artificial intelligence isn't likely to replace managers in the workplace, but it can help with routine tasks, a study by researchers in Carnegie Mellon University's School of Computer Science found. 

According to the researchers, AI manager clones — generative AI-powered digital surrogates trained on a manager's communication and decision-making patterns — could handle tedious or impersonal tasks in the workplace so human leaders can focus on mentorship, strategy and connecting with their employees.

They also found that employees often value managers for their emotional intelligence and personal understanding.

"Through this study, we learned that a lot of managerial work is informational and decision-based, and we've witnessed that AI is good at taking over these tasks," said Qing Hu, who led the study as an intern in the Collective AI Research and Evaluation (CARE) Lab in the Human-Computer Interaction Institute (HCII) and is now a master's student in CMU's School of Design. "However, when speaking with managers and workers, we found that management is also deeply interpersonal. Workers trust their managers, not just because of their decision-making abilities, but also because of the stories and experiences they share. I think what we are likely to see is not the manager disappearing, but their roles shifting."

Some workplaces are already introducing AI manager clones, and their adoption is expected to grow as technology advances. AI manager clones analyze emails, messages and meeting notes to learn how a supervisor communicates and makes decisions. The clone agent can join meetings, answer employees' questions, draft reports, analyze budgets and perform various other tasks. If a manager can't attend a meeting, the clone can join on their behalf.

The researchers interviewed 23 participants, comprising both managers and workers. In each session, they sought to understand how "AI bosses" can transform collaboration, power dynamics and organizational culture.

Participants questioned whether AI systems could ever fully replicate the nuance of human relationships, or if their presence might subtly erode them.

"Our participants expressed mixed sentiments about AI manager clones in the workplace," said Hong Shen, an assistant professor in the HCII and head of the CARE Lab. "While the technology could improve accessibility to busy managers and offers promising use cases, it also raises concerns about power dynamics and workplace relationships. From a human-computer interaction perspective, the findings point to critical design questions: how do we maximize benefits while minimizing risks so these systems are fair, transparent and truly support future work?"

On the practical side, participants recognized that AI manager clones can provide significant advantages. Managers are often overloaded with administrative tasks, and AI clones can provide more immediate responses. Employees might benefit from quicker feedback and consistent communication. AI can also help create a more equitable workplace, with reduced worker-manager hierarchies.

"If the inclusion of AI can diminish or even overturn power structures like leader-worker dynamics, then, as a worker, I'm willing to support anything that can facilitate this transformation," one of the study's participants noted. "In such visions, the absence of managers is not a loss, but a condition for greater happiness, with more room for individuals to exercise agency."

Study participants emphasized that managers should designate certain tasks as "human-only zones," including onboarding, conflict navigation and performance feedback. Workers also envisioned using agents to protect themselves, secure clarity or ensure commitments are honored. Specifically, AI manager clones might be used to clarify ambiguous expectations, document contradictory instructions and buffer emotionally difficult interactions.

"In our paper, we found some risks for workers," said HCII Ph.D. student Qing Xiao. "In the HCII, we place a strong emphasis on worker-centered design and technology for individuals who possess less power and authority. From the 23 participants we interviewed, we found the most resistance from workers, and we want to enhance their ability to maintain authority."

Another key insight from the study is that adoption will likely vary widely across organizations. More traditional workplaces might not trust AI management tools and will resist implementation, whereas technology-driven startups might embrace experimentation. Organizations that clearly communicate how AI tools operate — and where human oversight remains — will most likely find employees more willing to embrace the technology.

"From the business organization point of view, whenever we introduce some sort of technology to the workplace, the company's culture typically plays a role," said Hancheng Cao, an assistant professor at Emory University who collaborated with the CMU researchers on the work. "Some organizations are more conservative or traditional, so they might feel that certain AI use cases are not appropriate. But for other organizations, maybe a startup with tech-savvy individuals, this tool might be better utilized."

The researchers will present their findings at the upcoming Association for Computing Machinery Conference on Human Factors in Computing Systems (CHI 2026) conference in Barcelona.