AI programs fail more often on trust than on technology. How leaders talk about AI shapes whether people experiment, hide their use, or quietly resist.
What employees feel
BCG's AI at Work 2025 found that 41% of respondents worried their roles could disappear within the next decade, and that positive sentiment rises sharply when leaders actively support AI — yet only 25% of frontline staff report getting that support.
A cautionary tale
In April 2025 Duolingo's CEO sent an "AI-first" memo, later posted on LinkedIn, that included gradually stopping the use of contractors for work AI can handle and adding headcount only if a team cannot automate more of its work. It drew a strong public backlash. In August 2025 he told the New York Times: "This was on me. I did not give enough context," and said the company had never laid off full-time employees and did not plan to.
What good AI communication includes
- Why: the business reason, in plain words.
- What changes, what doesn't: tasks, tools, timelines.
- What it means for jobs: honestly — including what is not yet decided and when people will hear more.
- What support people get: training, time, redeployment paths.
- What stays human: the "human moments" list.
- How to ask and give feedback: named channels, regular pulses.
Culture habits that help
- Leaders using AI visibly and sharing what didn't work.
- Psychological safety to experiment and to raise concerns.
- Champions in each team; time set aside to learn.
- Recognition for improving work with AI, not for raw usage.
Sequence matters
Leaders first, then managers (with talking points and an FAQ), then all employees — with a pulse survey after 30 days. The AI policy kit generates this plan.