The 15-Minute Conversation
This is not a simulation. Schedule it on your real calendar with your real team. The structure below takes fifteen minutes and changes actual behavior.
3 minutes: Opening question
"When you think about using AI tools for your work, what feels useful, and what feels uncertain or risky to you?" Let people actually answer. Do not fill the silence.
5 minutes: Three ground rules
Share three non-negotiables:
1. No employee personal data in public AI tools, ever.
2. No AI-generated content shared externally without a human review pass.
3. No AI recommendation acted on without stating what human check was applied.
5 minutes: Agreement question
"What is one situation where we agree AI is clearly helpful, and one where we agree a human should always decide?" Capture both answers on a shared page. Two clear answers beat a long policy nobody remembers.
2 minutes: Closing commitment
"If you are ever unsure whether an AI use is appropriate, ask me before, not after." This one sentence does more than the rest of the conversation, because it makes uncertainty safe to surface.
Sam's story
Sam holds the fifteen-minute conversation with the team. The opening question surfaces something Sam did not know: two people were already pasting meeting notes with names into a public tool, not out of carelessness but because nobody had ever said not to. The ground rules fix it on the spot, without blame. The team agrees on one clear "AI is helpful here" (first drafts of routine docs) and one clear "human decides here" (anything about a person's rating or role). It took fifteen minutes and changed the team's actual behavior, which a policy document never had.
Knowledge Check
For each situation, select the right action.
A manager wants to draft a PIP for an employee who has been missing deadlines. They have detailed performance notes including the employee's full name and a personal health issue the employee disclosed.
What is the right approach?
A manager wants to summarize anonymous employee satisfaction survey data to identify themes for a team discussion.
What is the right approach?
A manager has interview notes for three promotion candidates and wants to use AI to synthesize which candidate performed best.
What is the right approach?
A manager wants to generate a set of structured interview questions for an upcoming software engineering hire.
What is the right approach?
A manager has a 45-minute client call recording and wants to summarize key decisions and action items.
What is the right approach?
Knowledge Check
1. The "noticeboard test" for AI privacy says:
2. AI-assisted people decisions can reproduce bias because:
3. Research on AI transparency shows that disclosing AI use:
4. Which is one of the three non-negotiable ground rules in the 15-minute team norms conversation?
5. Why is the manager described as the "accountability layer" for AI?
Apply It Today
Open your calendar right now and block fifteen minutes with your actual team this week or next. Use the four-part structure above. This is a real conversation, not a simulation.
Reflect