AI Is Polishing Your Hiring Process While Breaking Your Hiring Decisions

Most HR leaders buying AI for recruiting are solving the wrong problem. They're optimizing for speed and surface polish — faster screening, cleaner candidate communications, tighter scheduling — while the underlying machinery for making good hiring decisions is quietly degrading. The result is a process that looks more professional and performs less reliably.

That's not a hypothetical. New research flagged by HR Executive warns that AI-driven "knowledge decay" is actively eroding hiring trust. As AI tools mediate more steps between a candidate and a hiring manager, critical context gets lost or distorted at each handoff — what the researchers call a "risky game of telephone." The system produces confident-sounding outputs while the signal underneath deteriorates. You end up with a hiring decision that feels data-backed but is structurally less informed than the one your recruiter made manually three years ago.

The infrastructure problem compounds this. A ManpowerGroup Talent Solutions and Everest Group survey reported by HR Dive found that fragmented systems, isolated tools, and siloed data are blocking organizations from capturing AI's supposed recruiting gains. In other words: most companies are bolting AI onto a broken process architecture, then wondering why the outcomes don't improve. AI amplifies what's already there. If your foundational recruiting data is fragmented, you're not getting smarter decisions — you're getting faster bad ones.

Now add the legal exposure. A high-profile lawsuit against Workday has become the industry's clearest signal that AI hiring tools carry real discrimination liability, not theoretical risk. As HR Executive reported, one expert is explicit: AI use in hiring is creating serious legal risks, and the efficiency argument doesn't insulate organizations when disparate impact claims land in court. The question your general counsel is about to start asking — if they haven't already — is whether your AI vendor has been built for governance, not just performance. HR Executive's framing is direct: stop asking vendors "Does it use AI?" and start asking "Has the platform been built for governance?" That's the question that maps to your actual liability surface.

The governance gap connects to a broader AI ROI problem that should recalibrate board-level expectations. HR Dive's recent data roundup found that fewer than 5% of companies report "transformational" outcomes from AI investments. That number deserves to sit in front of every executive team that has approved an AI recruiting budget on the basis of vendor case studies. The 95% who aren't seeing transformation aren't necessarily deploying the wrong tools — many are deploying tools into organizations whose processes and governance structures were never redesigned to support them.

This is the structural misread most organizations are making: treating AI adoption as a procurement decision rather than an organizational redesign. MIT Sloan Management Review's work on AI ROI makes the case that measuring AI returns requires clarity about what kind of return you're even seeking — productivity, quality, risk reduction — and most companies haven't done that definitional work before signing contracts. In recruiting specifically, "efficiency" has become a proxy metric that obscures quality-of-hire and legal exposure, two variables that don't show up in time-to-fill dashboards.

The Workday lawsuit won't be the last. As AI embeds further into screening, scoring, and candidate ranking, the gap between "our AI vendor says it's compliant" and "we can demonstrate non-discriminatory outcomes in discovery" will widen for any organization that hasn't audited its own data inputs, model outputs, and decision audit trails. Efficiency without auditability isn't a feature — it's a deferred liability.

Senior HR leaders face a specific strategic choice here, not a technology tutorial. The organizations that will avoid the next wave of AI hiring litigation aren't the ones that move slowest — they're the ones that separated process redesign from tool procurement before the two got conflated. If your recruiting AI is running on top of fragmented data infrastructure, managed by a vendor whose governance architecture you haven't stress-tested, inside a legal environment where employment rights exposure is rising, you are not ahead of this problem.

The real question isn't whether AI belongs in your hiring process. It's whether your organization has done the harder, unglamorous work of fixing what AI will amplify.