A recent study by Boston University business professor Emma Wiles highlights that treating AI agents as coworkers can lead to poorer performance. The research found that managers made 18% fewer errors when they viewed AI as a tool rather than an employee. This raises concerns about the implications of AI integration in workplaces.
Understanding the Risks of AI as Coworkers
As companies increasingly frame AI agents as employees, the potential for mismanagement grows. Nearly a third of the 1,261 managers surveyed reported that their organizations treat AI agents as coworkers. This trend can lead to diminished accountability among human workers.
Wiles's study suggests that when AI is perceived as an employee, individuals feel less responsible for its output. Participants were 44% more likely to escalate questionable AI work to managers instead of trusting their own corrections. This behavior contradicts the intended efficiency of AI tools.
The Evolution of AI Agents in the Workplace
The rise of AI agents has sparked discussions among tech leaders about their role in the workplace. Jensen Huang, CEO of Nvidia, discussed the concept of “digital humans” last year, while major companies like Microsoft, OpenAI, Anthropic, and Google have developed tools aimed at managing AI teams.
While these AI systems have shown improved performance on complex tasks, referring to them as coworkers creates unrealistic expectations. This shift can result in confusion about responsibility and accountability, particularly in high-stakes environments such as healthcare and government.
Optimizing AI for Human Collaboration
Economist Daron Acemoglu, a Nobel Prize winner in 2024, argues that AI should enhance human capabilities rather than replace them. His perspective is echoed in a recent Stanford study that assessed the potential for AI to assist workers across various jobs.
Workers expressed interest in automation for specific tasks, such as law clerks wanting AI to track case progress. However, many tasks deemed suitable for AI by tech experts did not align with the needs of actual workers, indicating a disconnect in expectations.
- Study by Emma Wiles found 18% fewer errors with AI as a tool.
- Managers 44% more likely to escalate AI errors.
- Nearly 30% of managers treat AI agents as employees.
The branding of AI tools as employees, such as referring to them as “Alex,” does not enhance their effectiveness. Instead, it complicates the relationship between human workers and AI, ultimately hindering productivity.
🤖 This article was rewritten by Feed and Figures' editorial AI from a report originally published by MIT Tech Review AI. Facts and quotes are preserved from the original; the rewrite focuses on clarity and structure. For the unedited original, see the source link below.
