Lately, there has been a lot of talk about AI – artificial intelligence. As an avid reader of Sci-Fi, I take exception to the use of that term. These software applications may be artificial, but they aren’t intelligent. We are a long way away from R. Daneel Olivaw. But I digress. This week the Equal Employment Opportunity Commission released a New Resource on AI that is, to my mind, rather quite helpful.
Employers are warned to be careful in using AI when selecting employees. If the AI selection process has a disparate impact, tending to exclude protected classes, the employer had better be able to demonstrate job relatedness and business necessity. Fair enough, but what, exactly, qualifies as “artificial intelligence?” Dry, technical definitions of AI, such as the one provided by Congress,[i] don’t give me much comfort that I’ve correctly identified whether I’m using an AI application. This is where the EEOC’s New Resource comes to the rescue. It gives us concrete, tangible examples of AI. And by studying those examples, we are able to discern what is not AI.
The EEOC provides five examples:
- Resume scanners that prioritize applications using certain keywords;
- Employee monitoring software that rates employees on the basis of their keystrokes or other factors;
- “Virtual assistants” or “chatbots” that ask job candidates about their qualifications and reject those who do not meet pre-defined requirements;
- Video interviewing software that evaluates candidates based on their facial expressions and speech patterns; and
- Testing software that provides “job fit” scores for applicants or employees regarding their personalities, aptitudes, cognitive skills, or perceived “cultural fit” based on their performance on a game or on a more traditional test.
Notice that in all of these, the software application is gathering data – usually from the applicant directly (or their work product) – conducting an internal analysis of the data (often making a prediction), and then issuing a recommendation to the human user. If a software application doesn’t engage in these processes, then it probably isn’t AI.
Case in point: suppose an employer tells a background screening company to use its software application to identify all criminal convictions that involve the infliction of bodily harm on another person and to automatically classify all applicants with such convictions as “ineligible.” Is this AI? I would say it isn’t. The employer already made the selection decision in advance; it is simply using a computer to apply the criteria to a data set. The computer isn’t interacting with the applicant, it isn’t “interpreting” data supplied by applicant, and it isn’t making a recommendation to the employer on whether to hire. In short, the automated application of pre-defined criteria in this context doesn’t feel like the same thing a machine analyzing my facial expressions and then telling the employer whether I’m a good hire or not. (Yes, feelings are more relevant in the law than you might think.)
Of course, the conviction disqualification standard is still a selection process, and it is still subject to the EEOC’s Uniform Guidelines on Employee Selection Procedures (UGESS). So, this may be a distinction without a difference. No matter how smart, dumb, or artificial a selection process is, it is still subject to the UGESS.
[i] The National Artificial Intelligence Initiative Act of 2020 defined AI as a “machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments.”
This publication is provided only for educational purposes; it should not be relied upon as legal advice, and it should not be used, in whole or in part, as a basis for establishing practices or policies, nor should it be used for resolving disputes or managing risk. Every reader’s circumstance is unique, and legal advice should be obtained only from a lawyer with whom the reader has established an attorney-client relationship. Copyright 2023 © Higey Law LLC. All material contained within this publication is protected by copyright law and may not be reproduced without the express written consent of Higey Law.