AI Should Be Treated as a Tool, Not a Coworker

A study has found that marketing AI programs (agents) as 'digital employees' can reduce humans' ability to spot mistakes and make it easier to shift blame to AI when problems arise.

Major IT companies like Microsoft and OpenAI are releasing new tools focused on managing teams of AI agents, advertising them as 'digital colleagues' with flexibility and cognitive abilities akin to humans.

Research by Professor Emma Wilkins of Boston University found that when work outcomes were attributed to agent-style 'AI employees' rather than chatbots, people identified 18% fewer errors. The study indicated that when AI tools are perceived as employees, people feel less responsibility and are 44% more likely to report suspicious tasks to management.

Out of 1,261 participants in Professor Wilkins' study, approximately one-third of managers reported already considering AI agents as employees. 23% of participants even listed AI agents on their organizational charts.

These trends suggest that viewing AI as a colleague rather than simply a software tool can negatively impact human performance and accountability.