Beyond Checklists: Use AI to Surface Overlooked Talent and Catch Bias

The gist.

AI is changing how work and hiring happen across entry, mid, and senior levels. It can automate repetitive tasks and help teams make faster, better decisions. It is also being used to source, screen, and schedule candidates. The flip side is real. Algorithmic tools can narrow the funnel to people who match rigid keywords, and there are active lawsuits and enforcement actions focused on age, disability, and other protected traits.

Why it matters.

Box-checking misses people.
Research from Harvard Business School and Accenture shows automated filters routinely screen out qualified “hidden workers” who lack exact keywords or have non-linear careers, even when they could perform well with training.

Age bias risks are not theoretical.
The EEOC’s iTutorGroup case settled after applicants were auto-rejected based on age. A separate case against Workday’s screening tools has been allowed to proceed, with claims tied to race, age, and disability. AARP data shows older workers consistently report discrimination.

Disability and fairness rules apply.
EEOC and DOJ guidance warn that algorithmic tools can violate the ADA if they screen out people with disabilities without proper accommodations. There is no vendor liability shield. Regulators, including the FTC, DOJ, CFPB, and EEOC, have stated there is no “AI exemption” to civil rights law.

Compliance is tightening.
NYC Local Law 144 requires bias audits and candidate notices for automated employment decision tools. Colorado’s AI law treats hiring systems as high risk beginning in 2026, after a recent delay to June 30, 2026. More states are moving.

Getting started.

1. Map where AI helps and where humans must decide — List tasks by level. Automate rote steps, but require human review at decision points that affect people, with a written rationale. Use NIST’s AI Risk Management Framework to structure this.
2. Expand matching to find transferable talent — Stop strict keyword matching. Allow skills inference and adjacent experience. Calibrate your screeners against the Hidden Workers insights so nontraditional resumes are not silently excluded. Measure how many candidates advance after human review versus automated filters.
3. Install age-fairness and disability-fairness checks — Redact age proxies like graduation years where not job-related. Monitor adverse impact for 40+ candidates. Offer alternative, accessible assessments and clear accommodation paths to comply with the ADA.
4. Govern vendor tools like your own — Require bias audit summaries, explainability, and opt-out notices where applicable. In NYC, post-bias audit results and candidate notices. For Colorado, build toward “reasonable care” controls ahead of the 2026 start. Keep logs of model versions and overrides.
5. Track quality, not just speed — Pair time-to-first-interview and cost per hire with 90-day performance, hiring manager satisfaction, error rates in client work, and diversity mix. Use quarterly fairness reviews, not one-time audits.

Helpful links:

• Hidden Workers reports. Why keyword filters miss transferable talent. — Harvard Business School and Accenture
• EEOC iTutorGroup case. Age-based auto-rejection settlement. — EEOC
• Workday litigation. Court allows bias claims to proceed over AI screening. — Reuters
• ADA and AI hiring guidance. — EEOC and DOJ
• Joint regulators’ statement. — FTC, DOJ, CFPB, and EEOC
• NYC Local Law 144. Bias audit and notice requirements for automated decision tools. City guidance. — NYC Government
• Colorado AI Act. Hiring systems classified as high risk, effective June 30, 2026, after delay. State bill page and legal brief. — Colorado General Assembly
• AARP research. Prevalence of age discrimination concerns among workers 50+. — AARP
• NIST AI RMF. Practical structure for managing AI risk across the lifecycle. — NIST Publications