Why it matters now
People are already asking this question, whether or not their leaders are answering it. Silence fills with rumours, and careless announcements create a backlash that takes a long time to repair. I think leaders owe their people a straight answer, even when part of the answer is that some of it is not yet known.
- 41% of employees worry their role could disappear within a decade.BCG, AI at Work 2025 ↗
- After a backlash to the company’s “AI-first” announcement, Duolingo’s CEO said the reaction was “on me”.Quartz, Duolingo’s “AI-first” backlash, 2025 ↗
- Amazon’s CEO told staff that AI would mean fewer people in some jobs and more in others.Fortune, Andy Jassy’s memo to Amazon staff, 2025 ↗
Common mistakes
- Announcing “AI-first” without explaining what it means for people.
- Cutting roles before the AI capability actually works.
- Staying silent, which feeds fear and pushes AI use out of sight.
- Leaving role changes to attrition instead of designing them.
What to do instead
- Say what you know. Be honest that the mix of work will change. Say what you know, what you do not, and when you expect to know more.
- Pair efficiency with redeployment. Any plan to save time or cost comes with a plan for reskilling and moving people into new work.
- Redesign roles explicitly. Decide which tasks move to AI, which stay with people, and what new work appears. Write the new roles down.
- Prove the capability first. Change roles only once the new way of working has proved itself in real conditions.
- Share the gains. Give people a stake in what AI makes possible: time for better work, new skills, recognition or a share of the savings.
Your first step this week
Write the three sentences you would say if an employee asked you tomorrow what AI means for their job. Test them with a manager you trust, then say them to your team.
Related guides
- How do I get my people to actually use AI?
- How do we build AI skills across the organisation?
- How do we lead teams where people work alongside AI agents?
All AI leadership principles and guides · 100 AI use cases by function