Study of Dozens of Firms Finds Managers Miss More AI Errors
Updated
Updated · The New York Times · Jun 29
Study of Dozens of Firms Finds Managers Miss More AI Errors
1 articles · Updated · The New York Times · Jun 29
Summary
Researchers studying dozens of companies found managers checked documents less carefully when they were told an AI “employee” produced them, allowing mistakes to slip through that peers caught in human work.
Emma Wiles of Boston University and Boston Consulting Group collaborators traced the problem to a growing practice of treating AI agents as bona fide employees, sometimes even placing them on organizational charts.
Wiles said managers may see AI mistakes as someone else’s responsibility—belonging to tech teams or executives who deployed the systems—rather than part of their own oversight role.
The finding adds a subtler risk to already familiar AI problems such as bias, false answers and privacy leaks, suggesting companies may need to make managers explicitly accountable for AI subordinates.
When AI becomes a 'colleague,' why do managers lower their guard and miss critical errors?
Could the hidden cost of AI failures ultimately be 50 times greater than the initial savings?
The AI Accountability Crisis: Why 95% of Companies Fail to Realize Value from Generative AI
Overview
The rapid adoption of AI across many sectors has led to a growing crisis marked by widespread AI errors and failures in human oversight. Initial research highlighted AI's struggles with processing news content, and further studies confirmed these errors are systemic, affecting all languages and markets. This ongoing problem has prompted leaders to call on tech firms to focus on reducing errors, an area previously neglected. As a result, employees now spend significant time correcting AI-generated mistakes, which undermines productivity and highlights the urgent need for better oversight, robust governance, and a culture of accountability in organizations using AI.