Should Managers Use AI for Employee Problems?

Managers are already using AI for employee problems

Managers are already using AI to help them deal with employee problems. An employee keeps missing deadlines. Two people on the team can't seem to work together. A good employee wants a promotion but may not be ready. Someone's performance has slipped and the manager needs to address it.

These aren't new management problems. What's new is that a manager can now open an AI tool, explain what's happening and ask, "What should I do?" Within seconds, they'll probably have questions to ask, ideas for approaching the conversation and maybe even a script.

That can be very useful. It can also get a manager into trouble if nobody has explained where the boundaries are. My rule would be pretty simple: use AI to help you prepare. Don't use it to make the decision for you.

AI can help you prepare for an employee problem. But the employee is still your responsibility.

AI can be a very good sounding board

Managers have always needed somewhere to go when they're dealing with a difficult employee situation. Sometimes that's HR. Sometimes it's their own manager or another experienced leader. Sometimes they simply need to sit down and think through the problem before deciding what to do.

AI gives managers another way to do that. You can ask it to help organize your thoughts, suggest questions you haven't considered, practice a difficult conversation or even challenge your assumptions. Those are all reasonable uses.

The problem starts when we move from "Help me think through this situation" to "Tell me what I should do with this employee." Those are two very different things.

Prepare with AI. Decide with people.

Let's say an employee has missed several important deadlines. A manager could ask, "What questions should I ask before talking with an employee about repeated missed deadlines?" Or, "Help me organize the points I need to cover in this performance conversation." You could even have AI play the part of an employee who becomes defensive so you can practice how you might respond.

I like those uses because the manager is still doing the thinking.

Now compare that with, "I've told you everything about this employee. Should I fire her?" That's where I would stop. AI doesn't know the employee. It doesn't know what happened six months ago that the manager forgot to mention, whether expectations were clear, what coaching has already taken place or whether the manager contributed to the problem.

Most importantly, AI only knows the story the manager told it. That's not necessarily the whole story.

A better approach would be to ask, "What might I be missing?" or "What assumptions am I making?" Now you're using the technology to make yourself think rather than asking it to think for you.

Be very careful about what you put into AI

This is where organizations need to get very clear with managers. Employee problems usually involve information that doesn't belong in a public AI prompt. A frustrated manager could easily copy and paste an email exchange, performance review or complaint into an AI tool without stopping to think about what they're sharing.

Employee names and identifying information, compensation, disciplinary records, medical information, leave information, complaints, investigations and other confidential information need to be handled according to the organization's privacy, security and AI policies. Managers also shouldn't assume that because something can be discussed with HR, it can automatically be entered into an AI system.

If your organization allows AI for this type of work, take identifying information out whenever possible. Instead of entering an employee's name, age, location and detailed performance history, describe the problem without identifying the person. AI doesn't need the employee's life story to help you think.

Sometimes you need to stop prompting and call HR

There are situations where the right answer isn't another AI prompt. If you're dealing with discrimination, harassment, retaliation, accommodations, medical information, protected leave, workplace safety, threats, an investigation, whistleblower concerns, wage and hour issues or a possible termination, get the appropriate people involved.

AI might help a manager recognize that an issue needs to be escalated, but it isn't a substitute for HR or qualified legal advice.

There's another reason this matters. Imagine ten managers in the same company dealing with similar situations. Each one uses a different AI tool, explains the problem differently and gets different advice. Pretty soon you don't have one management practice. You have ten.

The organization still has to establish its policies, values and expectations. AI doesn't get to do that for you.

AI may agree with you when it shouldn't

How a manager describes an employee can change the entire conversation. Consider the difference between "My employee is resistant to change and constantly challenges my decisions" and "An experienced employee regularly raises concerns about proposed changes and asks for more information before we implement them."

That could be the exact same person.

The first description has already told AI what the manager thinks about the employee. If the manager's description includes frustration, assumptions or incomplete information, the answer may reflect those same problems.

That's why I'd encourage managers to ask AI to challenge their thinking: "What assumptions might I be making?" "What else could explain this behavior?" or "What information should I get before reaching a conclusion?" That's much more useful than asking AI to confirm something you've already decided.

You still have to know your people

People are complicated. A performance problem might actually be a training problem. The employee who supposedly has a communication problem may have a manager who isn't communicating very well either. Someone who looks unmotivated may not understand what's expected of them. An employee who doesn't appear ready for a bigger role may simply never have been given an opportunity to show what they can do.

AI won't magically know any of that. Someone has to talk to the employee, pay attention and listen.

That's the manager's job.

The same thing applies to succession planning

I think AI has tremendous potential in succession planning. It can help organize information, create competency models, develop questions, identify development needs and help leaders have better succession discussions.

Where I would draw the line is asking AI, "Who should be our next leader?"

There are too many things involved in that decision. Performance matters. Potential matters. Experience matters. So do relationships, judgment, aspirations, organizational culture and what the business is going to need next.

AI can help organize the information and the conversation. Leadership still owns the decision.

So what should organizations do?

I don't think the answer is telling managers they can't use AI. They're going to use it. The better answer is teaching them how to use it responsibly.

Organizations need to answer three basic questions: What can managers use AI for? What information can they give it? And when does AI stop and a person take over?

Preparation, brainstorming, learning, practicing conversations and questioning assumptions may all be appropriate uses. Managers also need clear rules about employee information, confidential and proprietary information, and which AI systems are approved. Finally, they need to know when it's time for HR, legal counsel, senior leadership or their own professional judgment to take over.

We don't need managers who blindly trust AI, and we don't need managers who are afraid of it. We need managers who understand what it's good for, recognize where it can get them into trouble and know when to close the laptop and deal with the person sitting across from them.

AI can help you prepare for an employee problem. But the employee is still your responsibility.

Sources

National Institute of Standards and Technology (NIST). Artificial Intelligence Risk Management Framework (AI RMF 1.0). https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-ai-rmf-10

National Institute of Standards and Technology (NIST). Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile. https://www.nist.gov/publications/artificial-intelligence-risk-management-framework-generative-artificial-intelligence

U.S. Equal Employment Opportunity Commission (EEOC). Employment Discrimination and AI for Workers. https://www.eeoc.gov/sites/default/files/2024-04/20240429_Employment%20Discrimination%20and%20AI%20for%20Workers.pdf

U.S. Equal Employment Opportunity Commission (EEOC). EEOC Hearing Explores Potential Benefits and Harms of Artificial Intelligence and Other Automated Systems in Employment Decisions. https://www.eeoc.gov/newsroom/eeoc-hearing-explores-potential-benefits-and-harms-artificial-intelligence-and-other