Why it matters now
Some well-known companies now expect AI use as a baseline and include it in reviews. I understand the intent: end the debate and get everyone moving. But a mandate without support breeds resentment, and measuring usage invites people to perform it rather than benefit from it. The expectation should be about better work, not more prompts.
- In 2025, Shopify’s CEO made “reflexive AI usage” a baseline expectation, included it in performance reviews, and asked teams to show why AI could not do the work before requesting more headcount.Analytics India Magazine, Shopify AI memo, 2025 ↗
Common mistakes
- Measuring usage counts and seats instead of outcomes, which invites performative use.
- Mandating AI without providing tools, training or time.
- Prescribing exactly how every team must use it.
- Rewarding visible enthusiasm rather than better results.
What to do instead
- Set the expectation clearly. Say that you expect people to try AI in their work, and explain why.
- Pair it with support. Provide approved tools, training and protected time before you expect anything.
- Assess outcomes, not activity. In reviews, look at the quality and impact of the work, and how people improved it, not how often they logged in.
- Let teams own the how. Teams know their work best. Let them decide where AI helps and where it does not.
- Share good examples. Regularly show real uses that improved the work, including small ones.
Your first step this week
Before setting any mandate, check three things for one team: do they have an approved tool, have they had hands-on training, and do they have time to learn? Fix whichever is missing first.
Related guides
- How do I get my people to actually use AI?
- Why aren’t we seeing a return on AI, and how do we measure it?
- How do we build AI skills across the organisation?
All AI leadership principles and guides · 100 AI use cases by function