SHRM26 Debrief · AI, Data & Tech · #1878

AI & HR: Enhancing Opportunities Responsibly in a New Era of Work

with Camille Olson

▶ Watch on the SHRM portal ~68 min, distilled

AI governanceemployment discrimination lawdisparate impact

"We didn't deploy a bad tool. We deployed the tool badly."

What it was about

AI adoption in HR is accelerating far faster than governance can keep up, and the law governing it is a patchwork of state and local rules, not federal, built around six recurring legal principles. HR must build shared vocabulary, catalog every AI touchpoint, and layer in human oversight to avoid liability while building employee trust.

By the numbers

209% each year

SHRM's reported rate of AI adoption acceleration in HR, described as the steepest adoption curve of any HR technology

11%

Share of organizations reporting mature, governed AI systems and deployment

14,000

Number of employees/applicants who have opted into the nationwide ADEA collective action in the Mobley v. Workday case

Key notes

Build a compliance matrix that maps every AI tool you use against each jurisdiction's specific legal definition (ADS, ADMT, AEDT, etc.) rather than relying on one summarized term.
Create an internal AI glossary and put it in your AI policy, vendor RFPs, contracts, and training materials so everyone uses consistent terminology.
Inventory every AI touchpoint across the full employee life cycle (sourcing through offboarding) because you can't govern what you haven't cataloged.

The contrarian takeThe current administration takes the position that disparate impact liability theory is unconstitutional, arguing the 1991 Title VII amendment codifying it conflicts with the Constitution, and has directed the DOJ to review state AI anti-discrimination laws that rely on it. Olson says that stance already caused Colorado to roll back its AI law.

Take this back Monday

Do this for your team

List every AI tool touching hiring, reviews, or promotions, then require a documented human sign-off before any AI-influenced decision is finalized.

Say this in your next leadership meeting

AI adoption in HR is growing 209% a year, but only 11% of us have governance mature enough to back it up.

Watch out for

Trusting a vendor's claim that a tool is 'bias mitigated' without obtaining and reviewing the actual audit ('vendor puffery' is not a legal defense).
Letting Gen AI hallucinate facts into performance reviews, PIPs, or termination memos that become official, discoverable employment records.
Treating AI governance as a one-time, 'set it and forget it' audit rather than an ongoing early/middle/periodic re-audit process since models continually change.

Fun fact · Camille Olson

She's testified before Congress and the EEOC itself on federal employment law, including the Paycheck Fairness Act.

If this landed, see these

↳ Go deeperPractical AI for HR: Real-World Examples in Microsoft 365 and SharePointTurns the governance mandate into concrete Purview controls inside actual Microsoft 365 tools.⇄ The counterpointStrategic AI Adoption: Designing Smarter WorkflowsComplicates the legal-risk framing: some AI layoffs are just business decisions in disguise.✦ The unexpected oneCulture Redefined: Architecting a Skills-Driven, Continuous Feedback CultureSame hallucination risk, different venue: AI faking sincerity in feedback instead of reviews.