SHRM26 Debrief · Legal & Compliant HR · #2580

Navigating the AI Hiring Maze: Decision-Ready HR in an Uncertain Landscape

with Dave Schmidt

& Bre Timko

& Niloy Ray

▶ Watch on the SHRM portal ~74 min, distilled

AI hiring complianceadverse impactvalidation evidence

"In this case, the analysis was that it would've been better to go with random chance and to flip a coin to decide whom to hire than to use this very expensive and prominent AI tool."

What it was about

AI hiring tools are subject to the same longstanding anti-discrimination and validation obligations as any selection procedure, plus a rapidly growing layer of state, local, and international AI-specific transparency rules. Organizations must proactively evaluate, validate, and govern these tools rather than wait to be sued.

By the numbers

62% of organizations

per SHRM research this year, are using AI somewhere in the organization

57%

of organizations in jurisdictions subject to state or local AI laws are unaware they are subject to those laws (SHRM research)

10-15%

of the AI hiring tool audits DCI Consulting has performed found vendors introducing unlawful bias mitigation strategies

Key notes

Classify any AI tool you're considering by type (predictive/deterministic, generative/probabilistic, or agentic) before evaluating it, because each type carries different risk and explainability profiles.
Treat any AI tool that screens people out or disadvantages some candidates over others as a selection procedure requiring adverse impact analysis and validation evidence, even if a human makes the final hiring decision.
Do not rely on vendor validation claims alone; demand evidence that validation was tied to actual on-the-job performance, not just performance in the selection process itself, and confirm it applies to your organization's jobs specifically.

The contrarian takeHuman-in-the-loop review does not eliminate legal risk or exempt an AI ranking/matching tool from being classified as a regulated 'selection procedure' — and in the case study presented, a costly, well-known AI resume-matching tool performed worse than flipping a coin because it failed to accurately parse resumes.

Take this back Monday

Do this for your team

Before renewing any AI hiring tool, demand written proof it was validated against real job performance for your specific roles, not vendor claims.

Say this in your next leadership meeting

AI hiring tools are selection procedures under the law, so we validate and document them proactively instead of trusting vendor assurances.

Watch out for

Assuming a human-in-the-loop review absolves the organization of selection-procedure obligations, when human review does not eliminate the underlying risk.
Accepting a vendor's validation claim at face value without checking whether it was validated against real job performance versus just against other people's judgments in the selection process.
Throwing all available data into an AI model ('the hopper') without vetting whether each data source is actually job-related, leading to contamination.

Fun fact · Bre Timko

Bre Timko is a non-practicing attorney with a J.D. in labor law alongside her I/O psychology Ph.D. — a rare dual-credentialed HR compliance expert.

If this landed, see these

↳ Go deeperDetecting Deception: Practical Skills for HR ProfessionalsSame evidence-over-intuition standard, applied to reading people instead of validating algorithms.⇄ The counterpointBuilding an AI-Enabled Recruiting TeamPushes the opposite instinct: stop being cautious and finally turn on the AI you already have.✦ The unexpected oneThe Chief Disruption Officer: Architecting Tomorrow's Workforce EcosystemThe same proactive-governance mandate, scaled up to HR owning enterprise-wide transformation.