Workers Better — And Organizations Worse

AI speeds up solo work, but can leave teams with a collaboration hangover

Recently, I watched a supply chain consultant, who is a self-confessed poor programmer, use AI to generate 7,000 lines of code in five minutes to predict demand across a supply chain. That is undeniably impressive on its face. But does that kind of individual productivity gain translate into better organizational performance over time? And, when AI gets embedded in teams, who and what are members learning?

My colleagues and I have studied these questions through interviews with everyone from junior developers to CIOs, accompanied by simulation modeling. We’ve found that AI can create considerable “technical debt”: shortcuts that make today’s work easier but tomorrow’s work harder. Extended projects, particularly those involving legacy code, last long enough for technical debt to accumulate that wipes out much of AI’s initial productivity benefit. There’s another wrinkle in this productivity paradox — organizations with less experienced workers should rely on AI tools less, not more. Less experienced workers get more help from AI initially, but they are less able to spot the technical debt problems it creates because they do not understand the broader context as well.

The team-learning story is different but leads to much the same place. With other colleagues, I’ve been modeling what happens when AI becomes part of collaborative work. Paradoxically again, AI can make individuals more capable while eroding the collective knowledge that makes teams work. One mechanism is that AI enables rapid specialization: Individuals get better at their own tasks so quickly that they lose the ability to understand each other’s work — or even who knows what. A second mechanism is that AI can take over tasks that once helped build and maintain shared knowledge. When that happens, the whole team can start forgetting how to work together without realizing it.

The crux of the problem is this: AI produces very visible individual benefits while some organizational costs remain largely invisible. AI can dramatically increase how much information someone can acquire and process, but it can degrade how to genuinely know, coach, trust, and coordinate with the unique and human others with whom we share problems and tasks. Organizations that confuse the two may become data-rich and relationship-poor.

Anderson, Edward, Geoffrey Parker, and Burcu Tan (2025). “The Hidden Cost of Coding With Generative AI.” Sloan Management Review, 67(1): 12-14.

Figge, P., Anderson, E., and Lewis, K. (2026). “AI-Human Learning Systems: Investigating the Strategic Role of AI for Organizational Learning.” Strategic Organization, 24(2), 307-342.