AI for HR and People Teams¶
In one line: Use AI to draft job descriptions, policies, onboarding material and survey summaries faster, but keep people decisions (who is hired, promoted, paid or let go) with humans, because employment is one of the most regulated places to use AI. Last reviewed: 2026-11-07
The short version¶
- Safe, high-value uses are drafting and summarising: job descriptions, interview guides, policy first drafts, onboarding checklists, answers to routine employee questions from your own handbook.
- The risky uses are the ones that score, rank or screen people. Anti-discrimination law still applies when a tool makes or shapes the decision, and the employer carries the responsibility, not only the vendor.
- Employee data is sensitive. Do not paste performance reviews, medical notes or investigation files into a tool your company has not approved. See the acceptable use policy.
- Always keep a human who can explain and reverse a decision, and keep a record of what the tool did.
- This page is a judgment guide, not legal advice. Rules differ by country, state and city, so check with employment counsel before you automate anything that affects a person's job.
Where it fits¶
| Task | AI helps with | You still own |
|---|---|---|
| Recruiting | Job ad drafts, plain-language rewrites, interview question banks tied to the role | Who gets interviewed and hired |
| Onboarding | Checklists, welcome material, a handbook Q&A assistant | What the policies actually say |
| Policy | First drafts, plain-English summaries of changes | Legal review and sign-off |
| Employee relations | Summarising your own meeting notes | Fact-finding, fairness, decisions |
| Surveys | Clustering open-text comments into themes | Deciding what is a real problem |
| Learning | Course outlines, quiz drafts | Whether the training works |
Copyable prompts¶
Job description rewrite
Rewrite this job description in plain English. Separate "must have" from
"nice to have". Remove jargon and any requirement that is not needed to do
the job on day one. Flag wording that could discourage some groups from
applying. Keep it under 400 words. Draft: [paste]
Structured interview guide
For the role of [title], write 6 behavioural interview questions and a
1-to-4 scoring guide for each, describing what a weak and a strong answer
looks like. Every candidate will get the same questions. Base the criteria
only on the job duties I list: [duties].
Survey themes (paste anonymised comments only)
Group these open-text survey comments into at most 8 themes. For each:
a name, a one-line summary, 3 representative quotes, and how many comments
fit. Say which comments you could not place.
Screening and ranking: the line to hold¶
- The US Equal Employment Opportunity Commission said in an October 28, 2021 announcement that AI tools used in employment decisions may "mask and perpetuate bias or create new discriminatory barriers to jobs," and that anti-discrimination laws still apply to them. It launched an initiative on AI and algorithmic fairness to help ensure compliance with the civil rights laws it enforces. (EEOC press release, read 2026-11-07.)
- Practical test: if a tool filters candidates out, ask the vendor how it was tested for different outcomes by group, and test it on your own applicant data. See bias and fairness testing.
- Some places add specific duties (bias audits, candidate notice). See US state AI laws. Local rules such as New York City's were not checked for this page.
Common mistakes¶
- Letting a resume-ranking tool run without ever looking at who it rejects.
- Pasting sensitive employee records into a consumer chatbot.
- Using AI to write a performance review from a few bullet points, then not checking it against facts.
- Assuming the vendor is responsible for compliance.
Next steps¶
- Acceptable use policy and choosing enterprise AI tools
- Responsible AI basics
- Privacy and data protection with AI
Sources¶
- US EEOC, "EEOC Launches Initiative on Artificial Intelligence and Algorithmic Fairness", October 28, 2021: https://www.eeoc.gov/newsroom/eeoc-launches-initiative-artificial-intelligence-and-algorithmic-fairness (fetched 2026-11-07; the page does not name Title VII or the ADA).