The hot topic in just about every field of endeavor these days is the degree to which Artificial Intelligence (“AI”) will affect it. But even before the explosion of AI in the workplace, employers have been using automated systems to assist in making employment decisions, especially at the hiring level. Employers and employees have both long been using systems like Indeed and Zip Recruiter, which have computer algorithms designed to match open positions with qualified candidates. These algorithms are hardly infallible, and it is almost certain that many candidates are passed over for reasons that have more to do with glitches in the algorithms than the qualifications of the candidates.

With AI, however, employers have the ability to tailor their job searches to more specific criteria, which could eliminate some of those glitches. But employers need to be careful about the algorithms they create using AI. Any algorithm that passes over individuals on the basis of protected classifications – which are quite numerous in 2026 – is obviously unlawful under Title VII and state and local anti-discrimination laws. Since 1971 and the Supreme Court case of Griggs v. Duke Power Co., 401 U.S. 424 (1971), it has also been unlawful to create conditions for hiring that have a disparate impact on those in protected classifications regardless of the employer’s intent. A hiring criterion that disparately impacts a protected set of applicants would be unlawful whether or not that is the intended aim of the employer. The Griggs case declared unlawful a hiring requirement of high school diplomas as having an unreasonably negative impact on African-American applicants for jobs that did not require that level of education. Griggs requires “the removal of artificial, arbitrary, and unnecessary barriers to employment when the barriers operate invidiously to discriminate on the basis of racial or other impermissible classification.”

Thus, if a company uses job descriptions in creating an AI program for screening applicants, it must be sure that the “essential functions” of the position are clearly separate from non-essential functions. A 50-pound lifting requirement might reasonably be considered essential to a package delivery position, but would not be “essential” to most receptionist jobs, even if it might be helpful when receiving deliveries. Instructing the AI program that the lifting requirement is essential to the receptionist job would likely have a disparate impact on disabled applicants. Similarly, an AI program that ranks applicants based on their colleges must be careful not to randomly rank historically black colleges and universities below others.

As employers wrestle with the best use of AI to strengthen their hiring and promotion processes, they must also keep in mind the potential legal pitfalls involved in the process.

For a printable PDF of this article, Click here.

THIS DOCUMENT PROVIDES A GENERAL SUMMARY AND IS FOR INFORMATIONAL/EDUCATIONAL PURPOSES ONLY. IT IS NOT INTENDED TO BE COMPREHENSIVE, NOR DOES IT CONSTITUTE LEGAL ADVICE. PLEASE CONSULT WITH COUNSEL BEFORE TAKING OR REFRAINING FROM TAKING ANY ACTION.