Problem
AI can support long solo work loops. That's fine on a solo project, but on a team, where learning from each other matters, it becomes a problem.
What Goes Wrong
It's easy to retreat into private work with the machine instead of collaborating with your team. You feel less embarrassed about mistakes, get help instantly, and can work whenever you want.
Consequences: less pairing, which means less learning, decreased code quality, fewer shared decisions, and weaker collective ownership.
Symptoms: hermit behavior, fewer discussions, hidden assumptions in the work, loneliness.
Example
This has always been an issue, but AI makes it easier to work alone, so we'll see more of it. It runs against the principles of agile software development, which are about people, feedback, and learning.
Solution
Bring AI output back to the team. Read code together. Discuss prompts, assumptions, and trade-offs.
