Is A. I. Biased?
06/09/2020
Long gone are the days where the terms “A.I.” or “Artificial Intelligence” seem like terms from the far future. The future is here. A.I. is being used in everything from the practice of law to medicine and the education of today’s children. Employers are even using it in order to determine who to hire into their companies.
At a past ABA Conference, there was actually a great panel discussion on A.I. in the workplace. Sure, arguably, its usage may make hiring more efficient, but is this usage biased? Is the A.I. biased? After all, something that is human-made is only as good as the humans who originally made it. Usage suggests that such tools have created situations that create undesirable results based upon classifications, such as race, sex, and more. In this author’s opinion, this does not lead to a problem, but rather a “challenge” that we should be addressing.
How Does A.I. Work?
A.I. has been defined differently by different people and organizations throughout the years. However, loosely speaking, a computer is making decisions that a human would have otherwise made. This capability primarily depends on (1) training, (2) data sets, and (3) repetition. A Medium article by Sherise Tan sets forth a good explanation. On a rudimentary level, we do this all the time as end-consumers. Did you like that song that Pandora just played? You give it a “thumbs up.” Hate it? Thumbs down. Amazon? Facebook ads? Same story. The programs learn from your choices and behavior.
Hire Me, Please
In the corporate hiring context, A.I. can be and has been used in multiple ways. One of which is filtering candidates for interviews. A popular method is scanning through résumés and application materials. In fact, this is what ZipRecruiter claims to do. However, things can sometimes backfire. Can it infer sexual orientation and religious beliefs due to social media information that may be obtainable? Perhaps there are biases created against schools that one may have attended?
Here’s My Résumé
At one point, Amazon created an A.I.-based program that was meant to review résumés in order to find “top talent.” In fact, it was similar to using Amazon’s website—candidates were given scores ranging from one to five stars. However, it turned out that the system was arguably biased against women. The program was trained to observe patterns in résumés that were previously submitted in the prior ten years. Of course, as a primarily technology company, Amazon has done what a lot of similar companies have done in the past couple decades—hiring more men than women. So, of course, somebody named “John” may automatically get a leg up over somebody named “Linda.”
This issue is not just Amazon’s alone, however. Kriti Sharma has a great Ted Talk on this same exact topic. If a hiring manager is looking for the “next tech leader of the company,” the program will pick up on the fact that the hiring manger is “hiring men.” According to Sharma, it is a simply a short leap from there from which the program understands that men may make better programmers than women. The original bias is reinforced.
They’re Interviewing Me!
Another example is actually using the technology in the interview process. The Wall Street Journal has a great video describing an example of this. At least one company called HireVue has already created software that uses what an interviewee says via video to analyze a multitude of things, including the interviewee’s micro-expressions and enthusiasm about the job. Unilever and Hilton are at least two companies that are definitely capitalizing on it. HireVue’s CEO, Kevin Parker, states that such technology is supposed to remove the human bias from the process. However, some are referring to this as “digital snake oil” and a “license to discriminate.” What makes a perfect employee anyways? Does somebody’s accent detract from their “enthusiasm” about the job? If English is not their native language, does that detract?
What Does the Law Say?
Has A.I. has become “above the law” because a machine is making the decision as opposed to a human? As most know, employers cannot discriminate on the basis of race, color, national origin, or religion. Nor can they—for example—generally discriminate based upon age or disability. Although not much has been yet stated regarding this new technology on the federal level, it is always prudent to prevent issues down the road. Therefore, if advising companies who create, implement, or seek to use such or similar A.I. technologies as already discussed, then one should make sure that the programs have been thoroughly vetted and tested before usage. Otherwise, such companies may have undesired lawsuits on their hands.
However, it may very well be such lawsuits that encourage lawmakers to pay attention to what is happening. As of now, although there is some movement on laws affecting A.I. in general, only Illinois seems to have taken it to the workplace. It has already approved legislation that seeks to toss in some transparency—Public Act 101-0260—otherwise known as the Artificial Intelligence Video Interview Act. It seeks protect the interviewees. The language is not extensive and consists of the following:
Section 5. Disclosure of the use of artificial intelligence analysis. An employer that asks applicants to record video interviews and uses an artificial intelligence analysis of the applicant-submitted videos shall do all of the following when considering applicants for positions based in Illinois before asking applicants to submit video interviews:
(1) Notify each applicant before the interview that artificial intelligence may be used to analyze the applicant's video interview and consider the applicant's fitness for the position.
(2) Provide each applicant with information before the interview explaining how the artificial intelligence works and what general types of characteristics it uses to evaluate applicants.
(3) Obtain, before the interview, consent from the applicant to be evaluated by the artificial intelligence program as described in the information provided.
An employer may not use artificial intelligence to evaluate applicants who have not consented to the use of artificial intelligence analysis.
Section 10. Sharing videos limited. An employer may not share applicant videos, except with persons whose expertise or technology is necessary in order to evaluate an applicant's fitness for a position.
Section 15. Destruction of videos. Upon request from the applicant, employers, within 30 days after receipt of the request, must delete an applicant's interviews and instruct any other persons who received copies of the applicant video interviews to also delete the videos, including all electronically generated backup copies. Any other such person shall comply with the employer's instructions.
Emoji claps for valid effort. However, as is already apparent, there may be issues that are not cured or are raised with this new legislation. For instance, what exactly is “artificial intelligence”? As was discussed before, there have been many definitions and iterations throughout the year in the scientific community itself; how has the law defined it? Second, how does this “consent” occur from the interviewee? Is it an adhesive contract or one that ends up being a non-descript paragraph within a Terms & Conditions of an application? Third, is there any recourse for those who might receive negative reactions due to withholding consent? These questions are just a start. Although it is a valid effort from the Illinois legislature, questions still abound.
For now, attorneys and companies alike should stay vigilant. As with any technology, it can be used a shield or a sword. Use it wisely.
Silvia Mansoor
Koussan Hamood PLC
Detroit, MI
silviamansoor.com