AI Screening Tools Under Scrutiny: Federal Courts Pre-certify Adea Collective Action

In a key decision by employers to use AI software in employment decisions, the U.S. District Court of California granted a group of job seekers who argued that they were rejected by the AI-based employment recommendation tool to a preliminary group certification under the Employment Age Discrimination Act (ADEA).[1] The court found that the claims based on the different effects of the algorithm's employment system were suitable for population treatment at least in this early stage. Currently, parties are engaged in the discovery of rank, and plaintiffs can provide notice of suits regarding similarly located individuals, participate in the suit and provide an opportunity to collectively hear their claims.

The court's decision shows what we believe is the first to believe in a country that precertifies collective action based on suspicion of age discrimination caused by the use of artificial intelligence in employment. As detailed below, this decision could have a significant impact on employers and vendors who rely on AI-driven recruitment technologies, ensuring increased compliance risk and improved scrutiny. Therefore, we encourage companies to develop and deploy AI tools at every stage of the hiring process to review practices regarding the way AI is being used and to minimize potential liability in this area.

AI-based tools and federal court litigation, Mobleyv. Workday, Inc.

Workday, Inc. provides cloud-based talent and applicant screening tools to employers nationwide. Its adoption products offer AIL and machine learning-based tools that allow you to evaluate applicants by scoring, ranking or recommending candidates based on resume data and employer preferences. These tools offer employers the potential to streamline their talent acquisition process.

However, Derek Mobley and four other plaintiffs argue that Workday's AI tools have strengthened bias among existing employers, resulting in them and other job seekers suffering discrimination by race, age and disability. Specifically, the plaintiffs argue that AI tools rely on mimicking (and exacerbating) pre-discriminatory employment practices that prefer biased training data or specific classes of job seekers. The plaintiffs also argue that in many cases the employer cannot proceed unless the applicant meets Workday's screening algorithm.

In line with his allegations, Mobley attempted to pre-certify collective action on his behalf.”[a]Workday, Inc. Applications for employment opportunities using the recruitment application platform of the company and was denied an employment recommendation. “Preliminary group authentication is the first step in a two-stage process of proofing collective action, paving the path of the parties to make lesson discoveries. As a result, the standard for obtaining preliminary collective certification in this first stage is low.

To fill this (usually low) bar, Mobley had to argue that he was similarly located in a potential location for the opt-in plaintiff. The court's decision focuses on an analysis that “locates similarly” and discovers sufficient similarity there, and the entire group (comprising millions of estimated plaintiffs, which can proceed to discovery).

Court possession

The decision only addresses Mobley's claims about age discrimination, and centers on whether Mobley similarly claims that he similarly locates the plaintiffs of the proposed group. The court refused to reject Workday's request to Mobley because the parties had already made some discovery to produce inconsistent results in future cases. Instead, the court adhered to the 9th Circuit precedent by assessing the issues based on “substantial claims” of similarity of complaints.

Second, the court held that the proposed group was similarly located at Mobley. This is because the plausible complaint claims explained a uniform policy applicable to all job seekers who created a different impact. More specifically, the court held that Mobley argued that Workday's AI recommendation system mechanism could cause different effects among job seekers over the age of 40, and that all such applicants “will rise and fall together.”

Address Workday challenges

Workday has argued many challenges to preliminary collective certification, including that. (ii) Contrary to Mobley's argument, the underlying policy in question is not uniform. (iii) means that it cannot be as collective as “natural variation” between the proposed group member qualifications, the number of employment they applied, and the individual rejection rates.

First, it argues that Workday itself does not recommend hiring candidates, as AI recommendation systems cannot automatically deny candidates without employer participation. The court has determined Mobley's work day claim Participated Differential impact practices were sufficient to ensure preliminary collective certification. For example, the court highlighted two specific work day AI tools that grade and recommend applicants.

On the second point, Workday argued that a uniform policy would not apply to applications submitted through Workday, as employers could decide whether to use Workday's AI capabilities. However, the court found that the proposed group already reflects the limitation that the employer in question must have used the Workday AI feature in assessing individuals' candidacy. The court also rejected Workday's argument that individual AI features may have different impacts across different employers. Unit-level differences do not defeat Mobley's claims that the integrated policy can produce a different net effect that can be proven by evidence.

In a third point, the court determined that personal variation in application experience between groups did not overpower preliminary group certification. Specifically, the courts have stated that each member of the group that Mobley was proposed to Similarly Although it is located, it is rather to identify similar material in case resolution. However, the court highlighted that individual differences between group members may be related at the merit stage of the case.

Main takeout

Although preliminary certification is granted almost regularly, the decision remains prominent as the court's previous retention rejects Workday claims and discovers that if the tool acts as a gatekeeper in employment decisions, it may be directly liable under the federal anti-discrimination laws. It is also clear that the courts do not necessarily apply more stringent standards to preliminary certification of putative collectively challenging AI screening tools, where uniform algorithms are used across multiple broad pools of applicants. In particular, unlike different treatment cases, different shock cases rely on statistics and do not require the employer's intentions to be demonstrated. This underscores the need for businesses to use AI in their recruitment process.

(1) Auditing automated decision-making tools;

(2) track how these tools affect the recruitment process;

(3) Ensure human surveillance as necessary.

This and other cases will continue to be tracked in the algorithm adoption space. As this case shows, there is an evolving legal environment created by ML tools created by AI and ML tools. Look forward to it more!