Beneficial Friction: How Humans Can Work with AI, Not Just Against It

For years, designers and product leaders chased the goal of removing friction from their work. But with the advent of generative AI, they've created a tool that does all the work for them, leaving humans to do little more than verify.

This 'lopsided partnership' is not just a problem in itself, but also erodes human skills and judgment over time. Research has shown that frequent use of AI can lead to a decline in critical thinking scores, as well as an over-reliance on automated recommendations (Gerlich, 2025).

A study by the IBM Institute for Business Value found that organizations that prioritize their human operating model and pair it with strong change execution achieve up to 73% higher revenue growth compared to their peers.

So what's the solution? According to Renée Richardson Gosline of MIT Sloan, it's 'beneficial friction' - intentionally adding cognitive and procedural speed bumps to workflows to encourage humans to scrutinize AI output. This approach has been shown to counteract automation bias and skill erosion (Gosline, 2025).