New research from Boston University's Emma Wiles reveals that framing artificial intelligence (AI) agents as human-like “coworkers” significantly hinders human workers' ability to detect errors and encourages them to offload accountability. This finding emerges as major tech companies like Microsoft, OpenAI, Anthropic, and Google rapidly deploy AI agent management tools, prompting a critical re-evaluation of how businesses integrate these autonomous systems into their workforce.
Anthropomorphizing AI Leads to Oversight Lapses
Treating artificial intelligence (AI) agents as human colleagues, complete with names and job titles, can lead to a substantial decline in human oversight and accountability, according to recent academic findings. This approach, increasingly adopted by businesses, makes human employees less effective at identifying mistakes in AI-generated work.
Research Highlights Accountability Shift in AI Integration
Emma Wiles, an Assistant Professor at Boston University's Questrom School of Business and a Digital Fellow at the MIT Initiative on the Digital Economy, leads research into the evolving dynamics of AI in labor markets. Her work, conducted through the Human-AI Interaction in Recruiting & Employment (HIRE) Lab, investigates how AI reshapes hiring and workplace interactions. A large-scale experiment co-authored by Wiles, published in May 2026, demonstrated that when AI systems were perceived as human-like employees, participants caught 18% fewer errors compared to when the work was attributed to a generic chatbot. This phenomenon, known as anthropomorphism, where human qualities are ascribed to non-human entities, can lead to a misplaced sense of trust and a reduced critical distance from AI outputs. For more insights into AI agents, refer to amazon.com/what-is/ai-agents/" target="_blank" rel="noopener noreferrer">AWS's definition of AI agents.
Tech Giants Push 'Digital Humans' Amidst Cautionary Findings
The study, involving 1,261 managers, directors, and executives across HR and finance in the United States, Canada, and the European Union, revealed that nearly one-third (31%) of respondents already perceive AI as a teammate or employee. Furthermore, 23% indicated their companies formally list AI agents on organizational charts. This trend aligns with pronouncements from industry leaders, such as Nvidia CEO Jensen Huang, who in October 2025, envisioned future workplaces populated by “digital humans” alongside biological employees.
"So future workforces in enterprise will be a combination of humans and digital humans." — Jensen Huang, CEO, Nvidia
Since April 2026, major technology companies have intensified their efforts to integrate AI agents into enterprise workflows. These include:
- OpenAI: Launched Workspace Agents in April 2026, building on its Frontier enterprise platform.
- Anthropic: Released Claude Cowork + Skills for general availability on April 9, 2026, alongside a new Managed Agents layer.
- Google: Introduced Workspace Intelligence at Cloud Next 2026 on April 22, enhancing Gemini agents with shared context across various applications.
- Microsoft: Continues to expand its comprehensive AI agent ecosystem, including Microsoft Copilot integrated across Microsoft 365.
These AI agents are autonomous software systems designed to perceive their environment, make decisions, and execute tasks independently to achieve specific goals. They operate in a continuous loop of perceiving, reasoning, acting, and learning.
What This Means
The implications of anthropomorphizing AI extend beyond mere semantics. When humans attribute agency and personhood to AI, it can inadvertently reduce their own sense of responsibility. This shift can lead to a dangerous accountability gap, where errors are less likely to be detected and the ultimate responsibility for AI-driven outcomes becomes blurred. Instead of treating AI as a human substitute, organizations should focus on redesigning workflows and governance structures. The goal should be to clearly define human roles in supervising and validating AI-generated work, ensuring that human accountability remains paramount. This approach fosters effective human-AI collaboration without compromising quality or ethical standards. Further research on the psychological impacts of human-AI interaction is crucial, such as studies exploring the effects of anthropomorphism on trust and responsibility.
Key Points
- Boston University research found people caught 18% fewer errors when work was attributed to an "AI employee" versus a chatbot.
- Nearly one-third of surveyed managers (31%) already frame AI agents as employees or teammates.
- Major tech companies like Microsoft, OpenAI, Anthropic, and Google have released new AI agent management tools since April 2026.
- Nvidia CEO Jensen Huang predicted future workforces will combine humans and "digital humans."
- Anthropomorphizing AI can reduce human accountability and lower quality control.
The Bottom Line
As AI agents become more sophisticated and integrated into daily operations, businesses must resist the urge to simply label them as "coworkers." The evidence suggests that such framing can undermine critical human oversight and accountability, potentially leading to increased errors. Companies should instead focus on clear governance, defining AI as a powerful tool, and establishing robust human-in-the-loop processes to maintain quality and responsibility in the evolving hybrid workforce. The future of work demands thoughtful integration, not anthropomorphic shortcuts.