The good and the ugly
Most teams use AI in private. The wins stay private, the misfires definitely stay private, and nobody builds a shared picture of what AI actually does and doesn't do well. Over time that means the team's collective judgment about AI never develops. This experiment makes those interactions visible, without it becoming a debrief or a debate. Try it when AI use is common but unspoken, or when you sense people are either over-trusting or quietly embarrassed by how they use it.
How to run it
Set up a dedicated channel for sharing AI interactions, good and bad. No commentary required, just share and let the examples speak.
Create a channel or thread for this, something like #ai-good-and-ugly or add it to an existing team space.
Agree on a loose norm: share one AI interaction a week each. Could be something that impressed you, surprised you, or went embarrassingly wrong.
React and respond lightly. The goal is noticing, not debating. A laugh or a "same" is enough.
If your team works in person or does regular team sessions, bring a handful of examples into a meeting occasionally for a short live look. 10 minutes, no agenda beyond sharing.
Recommended time
6 weeks.
What to notice
Is there a shift in how much people verify AI output before using it?
Are people becoming more selective about which tasks they hand to AI?
Does the tone stay curious rather than cynical?
What this might move
Sign in to activate this experiment for your team.