Should AI Use Be Part of Your Performance Review?
As AI becomes part of everyday work, organizations are beginning to face a question that would have seemed premature not long ago: Should an employee's ability to use AI be part of their performance review?
On the surface, the answer seems straightforward. If AI can help employees work more efficiently, analyze information more effectively, and improve the quality of their work, organizations understandably want people to use it. Some companies are already experimenting with AI fluency as part of performance expectations.
But there is a risk in moving too quickly from encouraging AI adoption to measuring it. Using AI is not the outcome. The outcome is still the work.
The more useful question for Performance Management is not whether an employee uses AI. It is whether that employee can produce strong outcomes in an AI-enabled environment. That requires more than knowing how to prompt a tool. It requires judgment, validation, subject-matter expertise, and an understanding of when AI adds value and when it does not.
AI Usage Is a Poor Performance Measure
Organizations like measures because they make expectations feel concrete. It can be tempting to track how often employees use an approved AI platform, how many prompts they submit, how many workflows they automate, or how frequently they interact with an AI assistant.
Those metrics may be useful for understanding technology adoption. They tell us very little about employee performance.
An employee could use AI constantly and still produce inaccurate, mediocre, or unnecessarily complicated work. Another employee might use it selectively because they understand exactly where it adds value and where their own expertise produces a better result. The second employee may actually demonstrate greater AI maturity.
This is why organizations should be cautious about turning AI adoption metrics into performance metrics. If employees believe they are being evaluated on how often they use AI, they will have an incentive to use it whether or not it improves the work. At that point, the organization may be measuring compliance with an AI initiative rather than meaningful capability.
The goal should not be to reward employees for using AI. It should be to recognize employees who know how to use the resources available to them to produce better outcomes.
Good AI Performance Starts With Judgment
The employees who create the most value with AI will not necessarily be the people who use it most frequently. They will be the people who understand how to apply it thoughtfully.
That includes knowing:
When AI can accelerate the work
What information and context the AI needs
How to evaluate the quality of its output
When an answer needs to be challenged or corrected
What data should not be entered into an AI tool
When human expertise, judgment, or conversation matters more
When the right decision is not to use AI at all
Those capabilities are much closer to judgment than technology proficiency, and that distinction matters for Performance Management.
Organizations have spent years building competency models around capabilities such as critical thinking, decision-making, collaboration, business acumen, problem solving, and communication. AI does not make those capabilities less important. In many cases, it makes them more important.
An employee working with AI must still recognize whether an answer makes sense. They need enough subject-matter expertise to identify missing context and recognize when a seemingly confident AI response is incomplete or wrong. They also remain accountable for the work they ultimately deliver.
That is a much more meaningful expectation than simply saying an employee "uses AI effectively."
Evaluate Outcomes, Not Tool Activity
The best way to incorporate AI into Performance Management may be to avoid treating AI usage as a separate performance measure at all. Instead, organizations can evaluate whether employees are using the tools and resources available to them, including AI, to achieve stronger outcomes.
Depending on the role, that could mean using AI to:
Improve the quality or accuracy of work
Solve a business problem more effectively
Reduce unnecessary administrative effort
Identify insights that might otherwise have been missed
Improve a customer or employee experience
Simplify a process
Create more time for higher-value work
Develop and share a better way of working with colleagues
Those are meaningful contributions. The AI is the enabler, not the accomplishment.
Focusing on outcomes also avoids creating an artificial advantage for jobs where AI is easier to use. A software engineer, recruiter, financial analyst, HR business partner, nurse, manufacturing supervisor, and sales leader may all have very different opportunities to work with AI. Expecting the same amount or type of AI usage across those roles would make little sense.
Performance expectations should continue to reflect the work itself.
AI Fluency Should Look Different by Role
This is where organizations may need to resist a familiar Talent Management temptation: creating one enterprise-wide AI competency and adding it to every performance form.
There may be value in establishing a common principle around responsible and effective AI use. But what good AI-enabled performance looks like will likely vary substantially by role.
For an analyst, AI fluency might involve using AI to explore data, test hypotheses, and identify patterns while independently validating the conclusions. For a manager, it could mean understanding which work can appropriately be delegated to AI while continuing to exercise sound judgment over the final result.
For an HR professional, it might include evaluating AI-generated recommendations while recognizing where employee context, policy, or human judgment needs to override them. For an executive, it may involve understanding AI's business opportunities and risks well enough to make responsible investment, workforce, and operating-model decisions.
The underlying expectation may be consistent across the organization: use AI responsibly and effectively to improve the work. The behaviors demonstrating that expectation should not necessarily be identical.
Organizations may therefore get more value from embedding appropriate AI expectations into existing roles, goals, and competency frameworks than from creating a universal AI proficiency score.
Managers Need to Be Ready for This Too
There is another problem with adding AI to performance expectations too quickly: managers have to be capable of evaluating it.
If an employee says AI helped reduce a task from four hours to one, can the manager determine whether the resulting work is actually better? If an employee automates part of a process, does the manager understand the new risks or controls that may be required? If someone deliberately chooses not to use AI for a sensitive or complex situation, will the manager recognize that as sound judgment rather than resistance to change?
Employees cannot be expected to demonstrate mature AI behavior if their managers cannot recognize what mature AI behavior looks like.
That means organizations may need to develop manager capability before changing the performance form. Managers need enough understanding to coach employees on appropriate AI use, evaluate the quality of AI-enabled work, recognize poor judgment, and distinguish thoughtful use from simple adoption.
Without that preparation, organizations risk introducing a new performance expectation that different managers interpret in completely different ways. That would create the same consistency and calibration challenges Performance Management teams already work hard to address.
AI Changes What We Mean by Performance
There is an even bigger question underneath this conversation.
For most of the history of Performance Management, there has been a relatively straightforward relationship between an employee and their output. The employee performs the work, and the manager evaluates the result.
AI complicates that relationship. Increasingly, an outcome may be produced by a combination of employee + AI + data + process + judgment.
If an employee can use AI to produce twice as much work, should the performance expectation simply double? If AI performs much of the initial analysis but the employee identifies the right problem, evaluates the results, and makes the final decision, where did the value come from? If two employees produce similar outcomes but one creates an AI-enabled workflow that can be reused across the organization, should that contribution be recognized differently?
There are also development implications. If AI performs increasingly sophisticated portions of a job, organizations will eventually need to distinguish between strong AI-assisted performance and the employee's underlying capability. This becomes especially important when evaluating readiness for promotion, expanded responsibilities, or succession.
An employee may be highly effective at combining their expertise with AI. That is valuable. But organizations still need to understand which judgment and capabilities reside with the employee, particularly when making decisions about future roles.
These are not reasons to avoid incorporating AI into Performance Management. They are reasons to be much more thoughtful about how we do it.
Start With the Work, Not the Tool
Before adding AI expectations to performance reviews, organizations should answer four questions.
1. What outcome are we trying to improve?
Performance Management should still begin with what the employee is expected to accomplish. AI should support those outcomes rather than become an outcome in itself.
2. Where can AI legitimately improve that work?
Not every task should be automated or augmented simply because it can be. Organizations need to understand where AI improves quality, efficiency, insight, or experience and where it introduces unnecessary risk or complexity.
3. What human capability remains important?
Judgment, expertise, relationship-building, creativity, accountability, and decision-making may become even more important as AI performs more of the routine work. Organizations should be clear about which capabilities employees are still expected to develop and demonstrate themselves.
4. What evidence would demonstrate effective AI-enabled performance?
That evidence should focus on results and judgment rather than raw usage. It may include better outcomes, stronger quality, improved efficiency, thoughtful validation, responsible decision-making, or the ability to redesign work in ways that create broader value.
Only after answering those questions should an organization decide whether AI needs to appear explicitly in a goal, competency, development plan, or performance review.
Looking Ahead
AI will increasingly become a normal part of how employees work. Eventually, asking whether someone "uses AI" may sound a little like asking whether they use email, spreadsheets, or search. The technology itself will become less interesting than what people are able to accomplish with it.
That is where Performance Management should stay focused.
Organizations should absolutely expect employees to adapt as work changes. But the goal should not be to create employees who use AI as often as possible. It should be to develop employees who know how to use AI well: employees who recognize where it creates value, challenge it when necessary, validate what it produces, and know when human expertise still matters more.
In an AI-enabled workplace, good performance will not be defined by who uses AI the most. It will be defined by who uses the tools, knowledge, and judgment available to them to produce the best outcomes.
What do you think? Should AI fluency become an explicit part of the performance review, simply become part of what good performance looks like, or does the right answer depend on the role?