{ "source": "Cooked Index — occupational AI risk register", "page": "https://cookedindex.com/jobs/human-resources-managers/", "methodology": "https://cookedindex.com/methodology", "notice": "Verdicts are re-examined as evidence accumulates. Re-fetch before relying on this; the page above always carries the current score.", "scored_at": "2026-08-11", "model": "claude-opus-5", "occupation": { "title": "Human Resources Managers", "soc_code": "11-3121", "category": "Management", "us_employment": 220660, "median_annual_wage": 149280 }, "verdict": "EXPOSED", "risk_resistance": 44, "contested": false, "near_boundary": false, "dimensions": { "task_resistance": 8, "embodiment": 4, "liability_shield": 5, "trust_premium": 12, "judgment_accountability": 15 }, "reasoning": { "task_resistance": "An 8 reflects the split in the actual calendar: EEO-1 and VETS-4212 filings, FMLA eligibility letters, offer-letter generation, ATS resume ranking and handbook updates are already vendor-automated, while the irreducible hours are the ones you cannot do through a screen — a two-hour investigatory interview with a complainant who is crying, a grievance hearing under a collective bargaining agreement, a termination meeting where the employee threatens to sue.", "embodiment": "A 4 is right because the job is Workday, Teams, and a conference room — the only physical requirement is being bodily present as the witness in a termination or an I-9 inspection, and even that survived remote work.", "liability_shield": "A 5 rather than a 0 recognizes that SHRM-CP or SPHR is a hiring filter at most employers but confers no legal authority, and when a Title VII or FLSA claim lands, liability attaches to the employer and outside counsel, not to your certification — you are a fact witness, not a licensed signatory like a CPA or an attorney.", "trust_premium": "A 12 rather than 16 reflects that employees will disclose a hostile-work-environment complaint or a substance-abuse issue to a specific HR manager they trust and not to an ethics hotline, but executives treat the function as replaceable headcount and the relationship rarely follows you out the door the way a client book follows a partner.", "judgment_accountability": "A 15 is earned by the calls with no procedure behind them: whether a harassment finding is substantiated on he-said-she-said evidence, whether to accommodate under the ADA or claim undue hardship, whether a reduction-in-force selection list has disparate-impact exposure — each one is defensible or not in a deposition years later, with your name on the investigation file." }, "rationale": "A large slice of the HR manager's week — drafting job descriptions and policy handbooks, screening applicants, summarizing engagement survey data, tracking benefits enrollment, generating compliance reports — is text-and-screen work that current AI handles at usable quality, and HRIS vendors are shipping exactly those features. What holds is the accountable human layer: conducting harassment investigations, sitting in on terminations, negotiating with counsel and union reps, and deciding contested compensation and promotion calls where someone has to own the outcome in a deposition. No license protects the role, so the moat is judgment and organizational trust, not regulation.", "outlook": "Headcount in HR departments thins as administration and screening automate, and the surviving HR manager role concentrates on investigations, labor relations, and executive advisory — fewer seats, higher stakes per seat.", "what_would_raise_it": { "levers": [ { "dimension": "liability_shield", "change": "EU AI Act Annex III classifies employment/worker-management AI as high-risk, requiring a named deployer to ensure human oversight and keep logs; Colorado SB 24-205 and NYC Local Law 144 push in the same direction. If a US state statute or an EEOC enforcement guidance names a specific HR officer as the accountable signer on adverse-action decisions produced with AI assistance — the way a plan administrator signs an ERISA filing — the HR manager becomes the person who is personally exposed in litigation for the algorithm's output rather than an optional reviewer.", "plausibility": "already happening", "would_add": 5 }, { "dimension": "liability_shield", "change": "Court practice under Title VII/ADEA disparate-impact claims where a defendant must produce a human witness who can testify they reviewed and adopted the screening criteria; if judges begin excluding vendor-generated validation reports absent a testifying human decision-maker (analogous to Daubert treatment of black-box evidence), employers will formally assign that role.", "plausibility": "plausible", "would_add": 3 }, { "dimension": "judgment_accountability", "change": "Task-mix shift: as handbook drafting, req writing, survey summarization and benefits tracking are absorbed by HRIS copilots (Workday Illuminate, SAP Joule), what remains in the week is disproportionately harassment investigations, contested terminations, RIF selection, and union grievance handling — all deposition-exposed calls under ambiguity. The score rises because the low-accountability tier evaporates, not because the job gains anything.", "plausibility": "already happening", "would_add": 4 }, { "dimension": "task_resistance", "change": "Same two-tier shift: investigative interviewing, witness credibility assessment, and settlement negotiation with plaintiff's counsel are not text-summarization tasks. If the residual role concentrates there, measured resistance rises even as total headcount falls — note this is compatible with fewer HR managers overall.", "plausibility": "already happening", "would_add": 3 }, { "dimension": "trust_premium", "change": "Collective bargaining language requiring that discipline, discharge, and grievance decisions be made by a human management representative — CWA and WGA/SAG-AFTRA contracts already contain AI-decision limits; if similar clauses spread through public-sector and healthcare CBAs, employers are contractually barred from automating the decision seat.", "plausibility": "plausible", "would_add": 3 } ], "ceiling_note": "No licensure exists for HR management and none is on any board's agenda, so liability_shield is capped well below professions with a signing statute; SHRM/HRCI certifications carry no personal legal liability. The trust premium is employer-side and thin — buyers of HR services are cost centers, and no plausible mechanism makes an employer pay more for a human handbook. Realistic ceiling is in the high 50s, and it is reached by the routine tier disappearing rather than by the role expanding." }, "adjudication": { "method": "two independent runs agreed on the verdict", "outcome": "corroborated", "run_totals": [ 44, 48 ], "run_verdicts": [ "EXPOSED", "EXPOSED" ] }, "employment_history": { "points": [ { "y": 2017, "emp": 136310, "wage": 110120 }, { "y": 2018, "emp": 143580, "wage": 113300 }, { "y": 2019, "emp": 154800, "wage": 116720 }, { "y": 2020, "emp": 156600, "wage": 121220 }, { "y": 2021, "emp": 166530, "wage": 126230 }, { "y": 2022, "emp": 181360, "wage": 130000 }, { "y": 2023, "emp": 200600, "wage": 136350 }, { "y": 2024, "emp": 215520, "wage": 140030 }, { "y": 2025, "emp": 220660, "wage": 149280 } ], "from": 2017, "to": 2025, "change_pct": 61.9, "comparable_from": 2019, "spans_soc_revision": true }, "pivots": [], "license": "https://cookedindex.com/terms" }