HR's role in AI adoption
AI adoption is a people problem wearing a technology costume. The five jobs only HR can do, the mistakes to avoid, and a first 90 days.
In most organizations, AI arrived through IT. IT picked the tools, set the controls, and sent the announcement. Then everyone waited for the change.
It didn’t come, because tools don’t transform anything. People do, when their work, their skills, their incentives, and their trust change together. That is HR’s territory. AI adoption is a people problem wearing a technology costume, and the function best placed to lead it has mostly been watching from the side.
Humans with AI outperform either alone. Getting there is a design job, and it belongs to you.
The gap nobody owns
Ask who owns AI in your organization and you’ll usually get a name in IT or a newly minted “AI lead.” Ask who owns how people learn to use it, how roles change, how performance is measured when AI did part of the work, and how employees are protected from bad decisions made with it. The room goes quiet.
That silence is the opportunity. Organizations are adapting unevenly: a few teams racing ahead, most unsure what is allowed, and a growing gap between the two. Closing that gap is the most important thing a People team can do this year.
Five jobs only HR can do
1. Fluency for everyone
Access is not adoption. Most employees with a license use it rarely and badly, because nobody taught them how to brief it, verify it, or know when to stop. A fluency program, personalized to your tools and your work, raises the floor for the whole workforce and is the single highest-leverage move available. See the AI Fluency demo.
2. Redesign the work, not just the tools
When AI takes the expensive middle out of a job, the job changes. Some of the hours go to higher-value work; some disappear. Someone has to decide which, role by role, and make sure the people doing the work are part of that decision. That is job design, and it’s yours.
3. Hire and develop for a different workforce
Credentials and experience tell you less than they used to. Judgment, curiosity, and the ability to work with a capable machine tell you more. Your competency model, your interview process, and your development paths all need an update, and the longer it waits, the more hires are made against the old one.
4. Make performance and incentives honest
If people are rewarded for volume, AI will produce volume. If managers can’t tell the difference between work a person understood and work a model produced, performance reviews drift toward theater. HR sets the standards for what good work looks like, what gets disclosed, and what gets rewarded.
5. Hold the lines that don’t move
Decisions about people, sensitive data, bias, disclosure, and accountability. AI can assist in all of them and must decide in none of them. HR, with Legal, defines those lines in language the organization will actually follow, and makes sure the tools and the training respect them.
The mistakes to avoid
- Announcing a policy instead of building a capability. A policy nobody has been trained to apply is a document, not governance.
- Waiting for the strategy to be finished. The strategy is written by doing. Start with one team and one real workflow.
- Treating adoption as the goal. Logins are not outcomes. Measure what changed for the work.
- Letting fear set the pace. Employees are already using AI, with or without permission. Silence pushes it underground; enablement brings it into the light.
- Skipping your own team. The People team has to go first. You cannot credibly enable an organization you haven’t enabled yourselves.
The first 90 days
Weeks 1 to 2: find out where you are. Who is using what, for which tasks, with what results. Where the risk is. Who the natural super-users are. A short survey and ten conversations will tell you more than a dashboard.
Weeks 3 to 6: your team first. Get the People team fluent on your tools, working on your real tasks. Build one small thing together. Write the context pack: what a model needs to know to be useful here.
Weeks 7 to 10: one partner function. Pick a team that’s eager and a workflow that matters. Run a Build Shop with them. Capture what worked as a repeatable format.
Weeks 11 to 13: make the case. A one-page summary with a number in it, a proposed policy for the lines that don’t move, and a plan for the next two functions. Present it as HR’s program, because it is.
Where we fit
Every part of this is something you can do. Learn It gets your team fluent. Build It turns ideas into working tools. Enable It gives you the programs, formats, and coaching to lead it across the business. Start with whichever one is under the most pressure.
Keep reading