Benefits, Risks and Best Practices
On a Friday afternoon, a manager stares at his screen and does the math. Six employees plus six overdue performance reviews. An entire weekend gone.
Why not let AI draft the reviews?
According to a survey of 2,000 HR professionals cited by the Society for Human Resource Management, 13% of employers report using AI in their performance review process, and that figure likely understates informal use by managers experimenting on their own, https://bit.ly/4l0rQKN.
If you’re a manager flirting with AI’s seductive promise: faster drafts, smoother prose, less emotional strain, you might be tempted. But pause first. Yes, AI can help. It also creates risks that many managers haven’t thought through.
Benefits
Drafting a polished review from scratch can eat hours, even if you’ve taken notes all year long. If you funnel those into AI software, you’ll be rewarded with a polished draft within minutes.
Managers who struggle with “how to say it” can use AI to turn fragmented notes into structured feedback, clarified expectations, and clearly defined next steps. Clearer language means feedback that’s more likely to drive improvement.
Further, a Gartner survey found that an astounding ninety percent employee believe algorithmic feedback could be fairer than traditional manager written reviews, https://www.charterworks.com/how-companies-are-using-ai-in-performance-reviews-now/.
Some managers even use AI-powered voice simulations to practice delivering difficult feedback. Rehearsing with a chatbot helps them evaluate their tone, anticipate employee reactions, and practice different responses.
Risks
AI doesn’t know your employees. Unless you provide actual, detailed observations, AI can create generic-employee fiction that sounds good but isn’t. What happens if an employee asks you what one of the comments you borrowed from an AI template means? Will you be able to give a concrete answer or only a vague generality?
Your organization uses performance review to decide wages, bonuses, promotions, and layoffs. What happens if your fast-AI work results in factual errors slipping in? Think that won’t happen? AI focuses on fluent writing and can easily generate praise or criticism that “sounds good” but lacks accuracy. McKinsey’s 2025 State of AI research finds that generative AI adoption is growing but warns that organization must build in human oversight, https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai.
AI wording often comes across as impersonal and generic. Sophisticated employees may catch the scent of AI when they read their reviews. If your employee suspects a robot wrote their performance review, you have a bigger problem than grammar.
Or, what happens if you draft a performance review using an AI template, and later fire or layoff the employee with the review as a crucial piece of documentation, and the plaintiff attorney shows the court or jury that you’ve done a “cut and paste” from an AI template.
Managers who use AI risk breaching an employee’s confidentiality concerning health accommodations or private situations. That information can be retained, incorporated into training models, or vulnerable in a security breach.
Take-aways
If you plan using AI as a tool, start with your own draft. Write the facts first. The more detailed and factual your notes, the better the result. Detail specific achievements, performance standards, missed targets, and detailed improvement-oriented information. Use AI to edit or refine clarity, not to invent substance. Avoid picking and choosing from AI-generated documents offering “samples.”
AI needs to reflect what you provide, not invent substance. A performance review needs to be employee-centered with references to actual performance, team dynamics and forward-looking goals aligned with your department or organization’s strategy.
Most importantly, review every sentence in the review to verify accuracy and remove exaggerated language.
Above all, remember you’re accountable. That’s your name signing the review, not Claude’s. Performance reviews shape careers. They influence pay, promotion, and morale. AI may give you back your weekend. Just make sure it doesn’t take away your judgment or your credibility .
© 2025 Lynne Curry, PhD, SPHR, SHRM-SCP
If you found this article useful, you might find value in Managing for Accountability; https://bit.ly/3T3vww8 a text that outlines how to effectively manage a high-performing team.
Subscribing to the blog is easy
If you’d like to get 1 to 2 posts a week delivered to your inbox (and NO spam), just add your email address below. (I’ll never sell it.) I’m glad you’ve joined this vibrant blog. Thank you!
Lynn, this sentence “Use AI to edit or refine clarity, not to invent substance” is crucial. I use AI in this way. I write everything myself- emails, proposals, etc- and then ask AI to edit for clarity and brevity, retaining my tone and vocabulary. Then I read it again, and I make the final edit. I never let AI write for me, I just have it improve what I wrote.
totally agree
I like your approach! This makes sense!
Too often, people are using AI to do their thinking for them and then claiming it boosts their productivity. It does mean they maybe put out more product, but what’s the quality, the accuracy, and accountability? As Lynne points out, the manager knows the employees, AI doesn’t. The approach of writing your own views first, then asking AI for edition help, then reserving the content of the final text for yourself, and your own emendations is much better and more responsible. If you don’t know how to write a performance review, maybe ask AI for advice or go on YouTube and see what’s offered.