We all came to the same conclusion. Or did we?
Everyone's thinking looks the same
Several people prepare with the same AI, so supposedly independent views may share the same framing, assumptions, or blind spots. Individual outputs can improve while the range of thinking across the group narrows.
- Different people's documents share the same structure and phrases.
- The range of options has narrowed since the AI tools arrived.
- The AI's first draft stays the frame, even after heavy editing.
Who: the people preparing for the next decision. Ask before the meeting: "Did you use AI to prepare, and which tool?" Look at: the options people bring. Count: how many differ in a real way. You're probably looking at this pattern if most people used the same tool and the options barely differ. Probably not if people used AI and still brought different views.
The first answer sets the frame, where one view takes over during the meeting.
Give it one instruction. Make the strongest case this is wrong.
In one experiment, AI help made each writer's stories better but made the stories across writers more alike (Doshi and Hauser, 2024). Emerging. That many people using one model will make the same mistakes together is our reasoning, based on formal theory (Kleinberg and Raghavan, 2021).