AI for Managers: Where It Helps With People Work and Where It Creates Risk
How managers can use AI for job posts, interviews, one-on-ones and performance reviews, and where US employment law, bias and privacy make AI use risky.
Managers spend a large share of their week on writing and preparation: job descriptions, interview questions, meeting agendas, feedback, reviews, update emails. AI is genuinely good at that kind of work. The same tools become risky when they move from helping a manager think to making or shaping decisions about people. Under US law, employers are responsible for discriminatory outcomes whether a human or software produced them, and employees notice quickly when feedback feels machine-written. The skill is knowing which side of the line a task is on.
Where AI helps, with low risk
Writing job postings. AI can turn a list of responsibilities into a clear, well-structured posting. Ask it to flag jargon, unnecessary requirements and wording that might discourage qualified applicants, such as demanding a degree when the job does not require one. Then check that the essential functions are accurate, because those matter for disability accommodation decisions later.
Structured interview guides. Ask for job-related questions tied to specific competencies, with a simple scoring rubric. Structured interviews, where every candidate gets the same questions and is scored on the same scale, are widely recommended for fairness and consistency.
Preparing for one-on-ones. Summarize your own notes from the past month, organize talking points or role-play a difficult conversation before you have it.
Clarifying your writing. Turning a long update into a crisp team email, or making feedback more specific and actionable.
Onboarding materials. Drafting checklists, first-week schedules and FAQ documents from existing internal docs.
Policy questions. Getting a plain-English explanation of a company policy you paste in, before you confirm with HR.
Where the risk rises
| Use | Risk | Safer approach |
|---|---|---|
| Ranking or filtering résumés | Disparate impact on protected groups; screening out people with disabilities or employment gaps | Use only HR-approved tools that have been evaluated; keep human review of rejected candidates |
| Analyzing video interviews | Bias, accessibility problems, state notice and consent laws | Avoid unless HR and legal have approved the tool and process |
| Drafting performance reviews | Generic or inaccurate claims, confidentiality, inconsistency | Write the assessment yourself; use AI to edit wording only |
| Recommending pay, promotions or terminations | Discrimination liability, lack of explanation | Keep decisions human, documented and based on job-related criteria |
| Monitoring productivity | Privacy concerns, morale, laws on electronic monitoring in some states | Follow company policy; be transparent with employees |
| Summarizing complaints or investigations | Confidentiality, missed nuance, legal privilege | Keep out of general AI tools; follow HR and legal processes |
The legal landscape in brief
This is a summary for awareness, not legal advice.
- Federal anti-discrimination law applies regardless of the tool. Title VII of the Civil Rights Act, the Age Discrimination in Employment Act and the Americans with Disabilities Act all cover hiring, promotion, pay and firing decisions. If an AI tool screens out a disproportionate share of a protected group, the employer can be liable even if nobody intended it.
- Disability accommodation. Online assessments, timed tests and video analysis can disadvantage people with disabilities. Candidates must be able to request reasonable accommodations, and the process should tell them how.
- New York City requires employers using automated employment decision tools for hiring or promotion decisions to have the tool bias-audited, publish a summary of the audit and notify candidates.
- Illinois has a law requiring notice and consent when AI is used to analyze video interviews, and has amended its Human Rights Act to address AI in employment decisions.
- Colorado has passed a broader law on high-risk AI systems used in consequential decisions, including employment, with implementation timelines that have shifted.
- Other states and cities have proposed or passed rules on automated decisions, electronic monitoring and biometric data. Illinois's biometric privacy law, for instance, regulates collecting things like face geometry.
Requirements are changing, so treat any HR technology purchase as something to clear with legal and HR, and ask vendors for documentation of bias testing and how the tool reaches its outputs.
Writing feedback with AI, the right way
Performance feedback is where AI temptation is highest and where employees are most sensitive to it. A workable method:
- Collect evidence yourself. Specific projects, outcomes, examples of behavior, and the goals set at the start of the period.
- Write your own rough assessment in bullet points: strengths, areas to develop, rating rationale.
- Use AI to edit, not to judge. Ask it to make your points clearer, more specific and more balanced, and to flag vague or loaded language like "abrasive" or "not a culture fit."
- Remove anything you would not say in person. If a sentence sounds like boilerplate, rewrite it.
- Keep confidential details out of tools your employer has not approved for personnel information.
The result should sound like you, be grounded in facts and stand up if someone later asks how the rating was reached.
Questions to ask before adopting an AI HR tool
- What exactly does the tool decide or recommend, and can a human override it?
- What data does it use, and was it tested for bias across sex, race, ethnicity, age and disability?
- Can the vendor explain why it produced a given score?
- How do candidates or employees request an accommodation or an alternative process?
- Does it meet notice and audit requirements in every location where you hire?
- Who owns the data, how long is it kept and is it used to train the vendor's models?
Prompts that keep you on the safe side
A few examples of requests that use AI for preparation without handing it the judgment:
- "Here is the job description. List the essential functions and flag any requirement that does not seem necessary to do the job."
- "Write eight structured interview questions for this role, each tied to one competency, with a one-to-five scoring guide."
- "Here are my bullet-point notes on a project. Rewrite them as specific, behavior-based feedback. Do not add any claims that are not in my notes."
- "Review this paragraph of feedback for vague, subjective or potentially biased language and suggest neutral alternatives."
- "Help me plan a 30-minute one-on-one about missed deadlines: an opening, three questions to understand the causes and a way to agree on next steps."
Each keeps the facts and the decision with you, and uses the model for structure, clarity and a second look at your wording.
Keep the human part human
The parts of management that build trust are listening, judgment and accountability, and those do not delegate well. Let AI handle the drafting, the structuring and the preparation, so you have more time for the conversations themselves. When a decision affects someone's job or paycheck, make sure the reasoning is yours, the evidence is documented and you could explain it face to face.
Frequently asked questions
Can I use a chatbot to write an employee's performance review?
You can use it to organize your own notes and improve the wording, but the assessment itself should be yours and based on documented, job-related evidence. Avoid pasting confidential personnel details into tools your employer has not approved, and read every sentence before it goes in the employee's file.
Is it legal to use AI to screen job applicants in the US?
It can be, but federal anti-discrimination laws such as Title VII and the Americans with Disabilities Act apply to AI screening just as they do to human decisions. Some jurisdictions add specific requirements, such as New York City's bias audit and notice rules for automated employment decision tools. Involve HR and legal counsel before adopting any screening tool.
Should I tell my team when I use AI?
Being open about it builds trust, especially for anything that affects people directly, such as feedback, schedules or evaluations. Some laws require notice for certain uses, like automated hiring tools in New York City or AI analysis of video interviews in Illinois, so check your organization's policy as well.
