Cognitive scientist Edwin Hutchins spent years studying how naval navigation teams actually solve problems — and found that intelligence wasn't located in any single person's head. It was distributed across the crew, the instruments, the charts, the social routines of hand-offs and double-checks. He called this 'distributed cognition,' the idea that thinking happens in the system, not just in the thinker. What strikes me about this today is what it implies about teams that have quietly replaced several of those human nodes with AI. The system still functions — maybe better by measurable metrics — but the cognitive load that used to travel between people, and therefore build their individual capacity over time, now routes around them. Nobody notices the erosion because the output looks fine. The practical implication isn't to distrust AI collaboration. It's to audit where your team's thinking actually lives: which problems do people genuinely wrestle with together, versus which ones get piped through a tool before anyone has really engaged? The former builds something. The latter processes it.

Name one decision your team or group made recently where the AI-assisted version and the unassisted version would have required fundamentally different conversations — and which one you actually had.

Drawing from Distributed Cognition / Cognitive Systems Theory — Edwin Hutchins