Should Managers Use AI for Employee Problems?

Managers have always needed help with these situations

An employee is missing deadlines. Another has become increasingly difficult to work with. A high performer wants a promotion but doesn't appear ready. Two team members are in conflict. Someone needs to be told that their performance isn't meeting expectations.

Managers have always needed help navigating situations like these. What's different today is where they may go for that help.

Instead of calling HR, talking with another manager, or searching through a management book, a manager can open an AI tool and type: “I have an employee who isn't performing. What should I do?”

Within seconds, AI can provide questions to ask, conversation starters, coaching approaches, performance-improvement ideas, and even a script for the meeting. That can be tremendously useful. It can also create significant problems if managers don't understand where the boundaries should be.

AI can help a manager think through an employee problem. It should not become the manager making the employee decision.

Why this matters now

AI has made management advice available on demand. That's a fundamental change in leadership development. A new manager doesn't have to wait for the next training program to learn how to structure a difficult conversation. An experienced executive can use AI to challenge their thinking before addressing a complicated situation.

Used appropriately, AI can help managers organize their thoughts, prepare questions, consider alternative perspectives, practice difficult conversations, improve the clarity of written communication, create coaching frameworks, brainstorm development opportunities, and identify issues they may not have considered.

But there's an important distinction. AI can help a manager think through an employee problem. It should not become the manager making the employee decision.

That's where organizations need clear expectations. Without them, every manager effectively creates an individual AI policy simply by deciding what feels appropriate at the moment.

Preparation versus decision

One of the simplest ways to establish a boundary is to distinguish between using AI to prepare and using AI to decide.

Suppose a manager needs to address an employee who has repeatedly missed deadlines. The manager might appropriately ask: “Give me five questions I can use during a conversation about repeated missed deadlines that will help me understand the cause before discussing solutions.” That's preparation.

The manager could also ask: “Help me organize the points I should cover in a constructive performance conversation.” That's preparation. Or: “Role-play an employee who becomes defensive when receiving feedback so I can practice responding professionally.” Again, that's preparation.

Now consider a very different prompt: “Based on everything I've told you about this employee, should I fire her?” That's asking AI to make a judgment about a person.

AI doesn't know that employee. It doesn't know the entire employment history, organizational policies, conversations that weren't included in the prompt, applicable legal considerations, the manager's own contribution to the problem, or whether the information supplied was complete and unbiased. AI is responding to the manager's description of the situation — not necessarily the situation itself.

A useful management rule is: use AI to generate questions, not conclusions. Ask it to challenge your assumptions. Ask it what information you may be missing. Ask it for alternative explanations. Ask it how you might approach a conversation.

But managers should remain accountable for understanding the employee, evaluating the circumstances, involving the appropriate people, and making the decision.

What should never be entered into a prompt

This may be the most immediate issue organizations need to address. Employee problems frequently involve confidential or sensitive information.

A manager frustrated by a situation might be tempted to paste an entire email exchange, performance evaluation, complaint, disciplinary document, or employee history into an AI tool. That can create unnecessary risk.

Managers should follow their organization's AI and data-security policies and should never assume that information appropriate for an internal HR conversation is automatically appropriate to enter into an external AI system.

Depending upon organizational policy and the AI environment being used, information requiring particular caution may include employee names and identifying information, performance evaluations, disciplinary records, compensation information, medical or disability information, leave information, complaints and investigations, allegations of harassment or discrimination, customer or proprietary information, confidential business information, personally identifiable information, and information protected by contracts, laws, or company policy.

The safest approach is to discuss the management problem without identifying the employee whenever organizational policy permits AI use.

Instead of: “John Smith, our 52-year-old sales manager in Dallas, has received two poor performance evaluations…” a manager can frame the issue generically: “A manager has experienced declining performance over the past six months despite receiving coaching. What questions should I consider before determining the appropriate next step?”

The purpose is not to give AI everything you know. The purpose is to obtain useful assistance without unnecessarily exposing information about the employee.

When HR and legal must be in the room

Some employee situations shouldn't begin — or end — with an AI conversation. Organizations should clearly define circumstances in which managers need to stop experimenting with prompts and involve the appropriate professionals.

Those situations may include issues involving discrimination, harassment, retaliation, disability or accommodation, medical information, protected leave, workplace safety, threats or violence, employee investigations, whistleblower complaints, wage and hour issues, termination, significant disciplinary action, employment contracts, or other legally protected activity.

AI may help a manager recognize that a situation could require escalation. It should not replace qualified HR or legal guidance.

There's another reason HR needs to be involved: consistency. Imagine ten managers asking ten different AI systems how to handle similar employee problems. Each provides slightly different information and receives different recommendations. The organization can quickly end up with ten different management practices.

Leadership cannot outsource organizational consistency to an algorithm. Policies, values, employment practices, and management expectations still belong to the organization.

AI can also reinforce a manager's bias

Managers don't necessarily approach AI with neutral descriptions. Consider the difference between “my employee is resistant to change and constantly challenges my decisions” and “an experienced employee frequently raises concerns about proposed changes and asks for additional information before implementation.”

Those could describe the same person. But they frame the situation very differently.

AI works with the information and context it receives. If the manager's description contains assumptions, frustration, incomplete information, or bias, the resulting advice may reflect that framing.

This is why one of the most valuable prompts a manager can use may be: “What assumptions might I be making about this situation, and what additional information should I gather before reaching a conclusion?” That's AI supporting better leadership rather than replacing it.

Managers still need to know their people

There is an appealing simplicity to asking technology for an answer. People aren't simple.

An employee's performance problem might actually be a training problem. A communication problem might be a manager problem. A supposedly unmotivated employee might lack clarity about expectations. Someone who appears unready for advancement may simply never have been given the opportunity to demonstrate leadership.

AI can't discover those things unless someone discovers them first and provides the information. That requires conversation. It requires observation. It requires listening. And it requires managers who are genuinely engaged with their people.

The same principle applies to succession planning

These questions become even more important when organizations begin using AI to support leadership development and succession planning. AI can help create competency models, readiness frameworks, development plans, interview questions, knowledge-transfer plans, and structured succession discussions.

But organizations should be extremely cautious about asking AI to determine: “Who should be our next leader?”

Succession decisions involve performance, potential, experience, aspiration, organizational strategy, culture, relationships, judgment, development, and future business requirements. AI can help structure that conversation. Leadership must own the decision.

What to do next

Organizations don't need to prohibit managers from using AI to address management challenges. They need to teach managers how to use it well. That starts with establishing clear expectations around three questions.

What can managers use AI for? Preparation, brainstorming, learning, practice, questioning assumptions, and developing possible approaches may all be valuable uses.

What information can managers provide? Organizations need explicit rules regarding employee information, confidentiality, proprietary information, and approved AI systems.

What decisions require human review or escalation? Managers need to know when AI assistance ends and HR, legal, senior leadership, or professional judgment must take over.

The goal isn't to create managers who automatically trust AI. And it isn't to create managers who are afraid to use it. The goal is to develop leaders who know how to use AI, how to question AI, and when to put AI aside.

Because technology may help a manager prepare for an employee problem. The responsibility for leading the employee still belongs to the manager.