Reading your team

Within a team, attitudes towards AI are not uniform. Spotting them avoids two mistakes: betting everything on the enthusiasts, who are already moving fast, or trying to win over the sceptics with arguments, which puts their backs up.

• The curious try on their own. They become champions, provided they share what they find.
• The cautious wait and see. A real case, shown by a colleague, wins them over.
• The sceptics doubt the quality or the point. They are valuable for testing the limits and setting the rules.

Give the sceptics a checking role: rereading an assistant’s answers, looking for its errors, proposing guardrails. Their high standards improve how the tool is used, and they often become the best guarantors of quality.

Amy Edmondson showed that teams learn when their members feel safe to take a risk in front of others: asking a naive question, admitting a mistake, sharing a failed attempt. With AI, this safety is decisive. A first try often disappoints, and someone who does not dare say so gives up in silence.

1. Tell the story of one of your own failed attempts and what you learned from it.
2. Publicly thank whoever reports an AI error.
3. Never compare people with each other on their usage.
4. Say clearly that nobody will be assessed on how well they master the tool.