AI in employment decisions: Critical questions emerging from the Workday and Meta cases
Recent lawsuits such as Mobley v. Workday and Doe v. Meta are forcing a shift in how companies approach artificial intelligence in the workplace. While much attention centers on whether a machine or a human makes a final call, the legal reality is more complex. Employers must now address whether their algorithmic systems account for legally protected activities, including disability accommodations and medical leave.
The central issue in these cases involves the design of performance metrics. Metrics like attendance history, time-in-position, or project completion rates often appear neutral on the surface. However, when these data points penalize employees who utilize FMLA leave or ADA accommodations, the resulting decisions can lead to discrimination claims. Simply having a human manager review an AI suggestion is insufficient if that manager does not possess the tools or the instructions to challenge the underlying data.
Effective oversight requires two distinct actions. First, organizations must document that humans exercise independent judgment when using AI tools. A rubber-stamp approval process provides little protection in a courtroom. Managers should know the specific factors driving a recommendation and retain the authority to override them based on individual employee circumstances.
Second, companies must audit the input data for bias. If a performance management model uses metrics that implicitly punish employees for taking protected leave, the entire system is flawed. Leaders should map where algorithmic tools influence employment decisions and regularly test these models for disparate impact. Accountability for model transparency must also extend to third-party vendors.
The takeaway for HR and benefits leaders is clear. Using AI for workforce planning is an active operational choice that requires constant governance. Relying on an algorithm is not a defense for biased outcomes. Responsibility rests with the employer to ensure that technology aligns with existing employment rights and that human reviewers remain the ultimate authority in personnel decisions.