“As organizations push to accelerate AI adoption, many leaders are considering a simple idea: Stop treating AI as a tool and start treating it as an employee,” said a recent Harvard Business Review article, “Why You Shouldn’t Treat AI Agents Like Employees.” The HBR article was co-authored by Boston University professor Emma Wiles along with Matthew Kropp, Julie Bedard, Megan Hsu, and Lisa Krayer from the Boston Consulting Group, based on a research study published in a May, 2026 paper, “Putting AI on the Org Chart: Evidence on Delegation and Oversight.”
The emergence of agentic AI systems makes this framing increasingly plausible. Unlike earlier software tools, agentic systems can operate with greater autonomy, coordinate workflows, make limited decisions, and execute increasingly complex tasks. As a result, some organizations are beginning to describe AI systems as teammates, coworkers, and even employees.
In particular, a November 2025 report, “The Emerging Agentic Enterprise: How Leaders Must Navigate a New Age of AI,” noted that traditionally, tools automate tasks while people make decisions. But agentic AI increasingly blurs these familiar distinctions by behaving partly like software and partly like a human coworker.
This hybrid nature creates organizational challenges that traditional management frameworks were never designed to address.
AI Employees Enter the Workforce
Management logic assumes relatively clear distinctions between technologies and people: tools automate tasks and employees make decisions; software functions as capital and workers provide labor. Agentic AI increasingly combines several of these characteristics at once. It can automate work while also exercising limited autonomy, interacting with employees, and coordinating activities in ways that resemble human collaboration.
As organizations integrate these systems into workflows, they must rethink fundamental management processes — including governance, accountability, workforce planning, role design, performance management, and employee training.
“Organizations are formalizing the role of AI agents in a variety of ways,” noted the article. “For example, they are giving them names and job titles, listing them on an org or work chart, assigning them managers, or referring to them as teammates or employees.”
“Some of these framing choices are social and symbolic,” adds the article. “Giving AI a human name, for example, gives people a way to address the system and, in some cases, can make it seem more familiar or approachable. Others are governance choices that trigger a set of expectations about authority, accountability, and oversight that the organization applies to its other employees. Whatever the specific practice, each reflects the same broader impulse — moving AI from something employees use to something the organization recognizes as a member of the team in its own right.”
What’s the impact of humanizing AI?
Why does it matter whether AI systems are treated as organizational actors rather than tools, and whether managers oversee them like software, delegate to them like human subordinates, or treat them as something distinct?
To better understand the organizational consequences of this shift, the researchers surveyed 1,261 HR and finance managers. They found that 31% said their organizations already frame AI as a “teammate or employee,” while 23% reported that AI agents appear on organizational charts.
Participants were then asked to review workplace documents containing errors. The only difference across experimental groups was whether the document had supposedly been drafted by an AI tool, a human employee (“Alex”), or an AI employee (“ALEX-3”).
The findings revealed several important organizational risks associated with anthropomorphizing AI that require work and governance redesign.
Accountability becomes blurred. “When AI was framed as an employee rather than as a tool, personal accountability fell by 9 percentage points, while accountability attributed to the AI rose by 8 percentage points.” This is a serious concern because AI systems cannot actually bear responsibility for decisions or errors. Accountability must ultimately remain with humans. When responsibility becomes diffused across teams and AI systems, organizations risk creating environments where people feel less ownership over outcomes and become more willing to overlook problems in AI-generated work.
Escalation and the burden on others increase. The researchers also found that managers were significantly more likely to escalate work for additional review when AI was framed as an employee. Compared with the AI-tool framing, the AI-employee framing increased requests for additional review by 44%. This creates additional organizational costs through more review cycles, slower workflows, and greater managerial burden, potentially causing so-called AI brain fry. Ironically, anthropomorphizing AI may reduce rather than improve organizational efficiency.
Quality control declines. Perhaps most concerning, managers exposed to the AI-employee framing identified significantly fewer errors in documents they reviewed. Compared with the AI-tool framing, participants caught 18% fewer errors when AI was framed as an employee. Managers in the AI-employee group were more likely to miss inconsistencies, flawed logic, and factual inaccuracies. These findings suggest that treating AI more like a coworker may unintentionally reduce critical scrutiny of its outputs.
Professional identity and trust weaken. The study also found important psychological effects. Managers whose organizations framed AI as a teammate or employee were 13% more likely to express uncertainty about their professional identity, a 7% greater concern about job security, and a 10% lower level of trust in how AI would be used within the organization. These findings suggest that anthropomorphizing AI may unintentionally undermine employee trust and engagement rather than strengthen support for AI adoption.
Adoption does not meaningfully improve. One of the most striking findings is that treating AI as an employee did not significantly increase willingness to adopt AI tools. Despite common assumptions, managers exposed to AI-employee framing showed no meaningful increase in adoption intent compared with those exposed to AI-tool framing. In practice, adoption appeared to depend far more on managerial expectations and organizational support than on symbolic framing choices.
Redesigning Work for Responsible Human-AI Collaboration
The study argues that organizations need to redesign work and governance structures to support more effective human-AI collaboration. As AI systems increasingly handle execution tasks, human roles will shift toward supervision, judgment, relationship management, and handling ambiguity. This transition requires organizations to redefine workflows, clarify oversight responsibilities, and develop new managerial skills.
Organizations must also rethink accountability structures. Oversight capacity does not automatically scale simply because AI systems generate more output. Companies need explicit rules defining:
- what AI agents can do autonomously,
- when human review is required,
- who intervenes when problems arise,
- and who ultimately bears responsibility for outcomes.
The researchers also argue that organizations need new training programs to help employees manage AI systems effectively. Employees must understand not only what AI can do, but also when its outputs should be trusted, challenged, or overridden.
One of the paper’s most important insights is that AI systems should not simply be mapped one-for-one onto traditional human roles.
“The concept of ‘employee’ assumes bounded roles, finite human capacity, and a hierarchy in which work is delegated to someone less experienced or knowledgeable,” noted the article. “AI does not share these limits.”
A single AI agent may operate across multiple workflows simultaneously, while multiple agents may reshape portions of a single job. Treating AI systems simply as human substitutes risks limiting organizational innovation and underestimating how profoundly workflows may evolve.
At the same time, organizations must remain attentive to the human side of this transformation. Simply asking employees to use AI to do more without redesigning roles thoughtfully is unlikely to strengthen engagement or professional identity. “While 76% of executives believe employees feel enthusiastic about AI adoption, only 31% of individual contributors report the same.”
The central issue is not whether organizations should deploy AI agents. They clearly will. The more important question is whether companies can redesign work, governance, and accountability structures thoughtfully enough to capture AI’s benefits without weakening human judgment and responsibility.
Treating AI systems as employees may feel intuitively appealing in an era of increasingly autonomous agents. But anthropomorphizing AI can blur accountability, weaken oversight, and create confusion about the distinct roles humans and machines should play inside organizations.
“In an agentic system, how work is divided between humans and AI is a design choice,” said the HBR article in conclusion.”As AI improves productivity, it creates the opportunity to redeploy human effort toward higher-value activities where demand is growing. Realizing that opportunity depends on deliberate role design, ensuring that human effort is focused on work that requires judgment, creativity, and ownership — areas that both drive value and reinforce positive engagement.”
The organizations that benefit most from agentic AI will likely be those that resist simplistic analogies and instead design new operating models that combine the complementary strengths of humans and intelligent systems.
AI disclosure: I wrote the initial draft of this blog and used ChatGPT for editorial assistance. I remain responsible for the final content.