In what may be the first discrimination suit involving artificial intelligence (AI), the Equal Employment Opportunity Commission (EEOC) recently filed a notice of settlement and request for a consent decree in an AI discrimination case. The agency’s lawsuit alleged that a New York-based English language tutoring service programmed its AI hiring tools to automatically filter out qualified applicants over specific ages. Affected parties filed charges of age discrimination with the EEOC. The EEOC has warned employers for months about the risk of AI discrimination when using such automated employment decision tools. In May, the EEOC and other agencies released joint statements on AI and automated systems in hiring.
Background of the AI Discrimination Case
According to the EEOC’s lawsuit, the New York English tutoring company requested dates of birth from applicants during the online application process. Next, the company allegedly programmed its AI software to automatically reject female applicants aged 55 and older and male applicants aged 60 and older. This practice violates the Age Discrimination in Employment Act of 1967 (ADEA). Specifically, the ADEA prohibits employers from discriminating against employees and applicants 40 years of age and older. In this case, applicants were indeed qualified for the position but fell victim to AI discrimination.
One female applicant initially stated in her application that she was older than 55 and was subsequently rejected. However, when she resubmitted her application with a more recent date of birth, the company offered her an interview. The lawsuit states that the company rejected more than 200 similarly qualified applicants because of their ages.
AI Discrimination and EEOC Guidance
While using AI in hiring decisions can help with a variety of employment matters, there is a risk of discrimination. Using automated systems for evaluating and filtering candidates, monitoring employee performance, and determining pay or promotions can inadvertently discriminate against otherwise qualified candidates based on protected classes. Without proper independent audits and other safeguards, AI systems may risk violating federal civil rights laws. According to recent guidance from the EEOC, employers should do the following to avoid AI discrimination:
- Conduct self-analyses of AI tools on an ongoing basis to determine whether their employment practices have a disproportionately large negative effect on protected classes.
- Perform an independent bias audit on the AI tools they use.
- Take additional steps to reduce the adverse impact on protected classes or select a different tool in order to avoid discriminatory hiring practices.
Proposed Consent Decree and Settlement
In EEOC v. iTutorGroup, Inc., et al., the parties entered into a proposed consent decree and notice of settlement for $365,000. The proposed consent decree also would prevent the tutoring company from doing the following:
- rejecting qualified applicants age 40 and over because of their age;
- screening applicants based on age or requesting dates of birth before they make a job offer;
- rejecting any applicant based on their sex; and
- retaliating against employees engaged in protected activity.
Under the proposed consent decree, the company must also provide discrimination training to all supervisors and managers.
Proposed “No Robot Bosses Act of 2023” Currently in the Senate
In an effort to curb AI discrimination in employment, senators in Pennsylvania and Hawaii recently proposed the “No Robot Bosses Act of 2023.” The proposed bill would require employers to disclose that they use automated systems to make employment decisions and how they are using these tools. Additionally, the proposed bill would prohibit employers from exclusively relying on AI tools in hiring, require testing and validation before implementing AI, require training on such AI tools, and implement independent audits and reporting on AI decision-making tools.