Artificial Intelligence in the Hiring Process

Machine Learning and Artificial Intelligence in Analytics” by deepak pal is licensed under CC BY 2.0.

Blythe McGregor, Associate Member, University of Cincinnati Law Review

The interview stage of the hiring process traditionally involves human to human interaction and an applicant is judged on his or her interpersonal skills. Increasingly, employers are turning to video interviews, in lieu of a face to face interview, to expedite the hiring process and cut costs.[1] Sometimes it is not another human watching and critiquing these videos. Employers, using a video and artificial intelligence platform such as HireVue,[2] may choose to have applicants’ answers evaluated by artificial intelligence (“AI”). Platforms like Hirevue program machines to simulate human analysis of movement, word choice, and voice to examine the interviewee.[3] The computer program then gives each applicant a score on “employability,” although often, the AI analysis takes place only during the first round of interviews.[4] Large companies, such as Unilever and Hilton, use this technology on thousands of applicants during the hiring process.[5]

Although use of AI may reduce employers’ burdens of reviewing and scoring each video interview personally,[6] critics of this hiring method argue that it is not rooted in psychology and could discriminate against nervous or non-native speakers.[7]  Additionally, as with any system that encodes and stores large amounts of personal data, there are also privacy concerns.[8] The Illinois General Assembly responded to the latter class of concerns in early 2019 by proposing the Artificial Intelligence Video Interview Act (“AIVIA”).[9] AIVIA became effective on January 1, 2020.[10] AIVIA affects hiring processes for jobs that are based in Illinois.[11] Although Illinois was the first state to pass legislation regulating the use of AI in the interview process,[12] it is not the first AI-hesitant regulation to be enacted. Other states have enacted regulations on the use of AI in the context of the testing and use of autonomous vehicles.[13]

Artificial Intelligence Video Interview Act

AIVIA imposes five restrictions on employer use of AI to analyze video interviews.[14] First, Section 5 of AIVIA imposes three steps to disclose information about the AI program and obtain consent from the applicant.[15] Initially, the employer must notify the applicant that AI will be used to analyze and consider the applicant’s interview.[16] Next, employers must provide each applicant with information about how the particular artificial intelligence works and what types of characteristics it uses to evaluate applicants.[17] Finally, employers must obtain consent from the applicant to subject their video to the type of analysis identified in the previous step.[18] All three of these requirements must be satisfied prior to asking applicants to submit a video interview.[19] AI cannot be used to evaluate a nonconsenting applicant.[20] Importantly, AIVIA does not define “artificial intelligence” or indicate the depth of explanation required by the second step. This will be discussed further in the analysis section.

Section 10 of AIVIA limits an employer’s ability to share video interviews that have already been submitted.[21] Employers can only share videos with those whose “expertise or technology is necessary in order to evaluate an applicant’s fitness for a position.”[22] Finally, Section 15 limits how and when an employer can store the submitted video interviews.[23] If an applicant requests that an employer delete the video, the employer must comply with the request within 30 days.[24] The employer must also instruct anyone else who received the video to delete it along with any backup copies.[25]

Analysis

Although AIVIA requires enhanced transparency when employers use AI to evaluate interviews, the act does not provide definitions or detailed guidelines to direct employers in the implementation of these new policies. AIVIA does not define AI, so it is not clear exactly what methods of video evaluation are subject to these regulations. Employers likely use many computer programs to sort and analyze information about applicants’ interviews, even when a human is the one evaluating the video in the first place. How removed from direct human involvement must the process be to qualify as “artificial intelligence analysis” under AIVIA?

Also, there is some ambiguity regarding how much information an employer is required to give to fulfill the second disclosure requirement. An employer must explain “how the artificial intelligence works” and disclose the “general characteristics” it uses to evaluate.[26] However, as explained above, employers themselves are not programming and implementing the AI program. Employers are using programs designed and provided by companies like HireVue to evaluate applicants. Employers utilizing this program may not have a detailed understanding of how the AI technology is structured and operates and therefore, may not be able to give the technical explanation required under AIVIA. Also, although the employer may know the general characteristics the technology identifies, because the employer is not programming the technology themselves, the employer may not be aware of when the AI system flags more nuanced facial movements or mannerisms. The third-party AI provider could possibly provide this information to the employee, but it is unclear whether AIVIA requires this information to come directly from the employer.

The logistics of implementing these policies are also ambiguous. First, AIVIA does not specify in what form notice should be given to the applicant or how the applicant should give consent. This could create enforcement issues if there are no physical records to be used as evidence that the company has complied with AIVIA’s notice and consent requirements. There is also little guidance as to how videos should be deleted upon request from an applicant under Section 15. This Section is titled “Destruction of Videos” yet the statute simply calls for deleting the video. Merely deleting a file does not necessarily mean it is completely irretrievable from a computer’s hard drive.[27]  Next, AIVIA lacks complete privacy protection as it merely requires employers to “instruct” other persons who received the video to delete it and requires such other persons to comply with this instruction. Once again, enforcing this requirement seems implausible as it may apply to people or companies besides the employer, and it will be difficult to track whether they complied with deletion. Also, if an applicant does not give consent, AIVIA does not require employers to consider applicants using another method. Some applicants who have concerns about the use of AI, for example that the use of AI could lead to discriminatory results, may be disqualified from the job before they are even considered. AIVIA also does not identify what liability and damages an employer may incur upon violation of the act. With AIVIA recently taking effect, employers may not yet be incentivized to strictly comply. Because of these logistical ambiguities, applicants would not be remiss in thinking AIVIA does not completely protect their privacy.

Conclusion

AIVIA is a simplistic start to a possible trend toward increased regulation of AI in the hiring process. The act requires heightened employer transparency but, by language alone, does little to ensure that an applicant’s privacy is completely protected once they have submitted a video. Also, the act does not fully address concerns about whether the use of AI creates an unfair hiring process. Notably, AIVIA does not require employers to use only nondiscriminatory AI systems. Because the AI analysis turns on the mannerisms and speaking style of applicants, these systems may unintentionally discriminate against minority or non-English speaking applicants whose mannerisms and way of speaking differs from the baseline “ideal” applicant characteristics programmed into the system. Companies may look to successful current employees when identifying characteristics of an ideal applicant, and, if all these employees are men, for example, the interviews of those in a protected group may be systematically rejected.[28]  Additionally, applicants who do not consent to the use of such technology may not have another way to participate in interviewing, and thus, may consent to the use of AI despite legitimate concerns of discriminatory results.

Although the use of AI to evaluate applicants may save employers time, this efficiency comes at a cost. More detailed and comprehensive legislation is needed to protect job applicants from discrimination and invasion of privacy inherent in this new hiring technology. State legislation should regulate AI technology at its source: the third-party AI service providers. Strict regulation could reduce discriminatory impact and ensure the safe storing and sharing of video data. Legislatures may not yet have the full understanding of AI technology necessary to create and enforce this type of statute. As the use of AI technology becomes more widespread in other contexts and the dangers of its use become more apparent, extensive AI regulation will likely become more common.


[1] John W. Schoen, Lights, Camera, Job Interview!, CNBC (27 Jan 2014), https://www.cnbc.com/2014/01/24/shortcomings-evident-as-video-job-interviews-increase.html.

[2] Drew Harwell, A Face-Scanning Algorithm Increasingly Decides Whether You Deserve the Job, The Washington Post (Nov. 6, 2019), https://www.washingtonpost.com/technology/2019/10/22/ai-hiring-face-scanning-algorithm-increasingly-decides-whether-you-deserve-job/.

[3] Id.

[4] Id.

[5] Id.  

[6] Rebecca Heilweil, Artificial Intelligence Will Help Determine if You Get Your Next Job, Vox (Dec. 12, 2019), https://www.vox.com/recode/2019/12/12/20993665/artificial-intelligence-ai-job-screen.

[7] See Harwell, supra note 2.

[8] Tom Starner, AI Can Deliver Recruiting Rewards, But At What Risk?, HR Executive (Dec. 31, 2019), https://hrexecutive.com/ai-can-deliver-recruiting-rewards-but-at-what-legal-risk/.

[9] Artificial Intelligence Video Interview Act, Ill. Public Act 101-0260, http://www.ilga.gov/legislation/publicacts/fulltext.asp?Name=101-0260.

[10] Id.

[11] Employers Using AI in Hiring Take Note: Illinois’ Artificial Intelligence Video Interview Act is Now in Effect, Lexology (Feb. 10, 2020), https://www.lexology.com/library/detail.aspx?g=39ee6ae8-d114-47e3-a535-f6c721064252.

[12] Rebecca Heilweil, Illinois Says You Should Know if AI is Grading Your Online Job Interviews, Vox (Jan. 1, 2020), https://www.vox.com/recode/2020/1/1/21043000/artificial-intelligence-job-applications-illinios-video-interivew-act.

[13] Regulation of Artificial Intelligence: The Americas and the Caribbean, Library of Congress, https://www.loc.gov/law/help/artificial-intelligence/americas.php.

[14] Artificial Intelligence Video Interview Act, Ill. Public Act 101-0260, http://www.ilga.gov/legislation/publicacts/fulltext.asp?Name=101-0260.

[15] Id.

[16] Id.

[17] Id.

[18] Id.

[19] Id.

[20] Id.

[21] Id.

[22] Id.

[23] Id.

[24] Id.

[25] Id.

[26] Id.

[27] Are Deleted Files Completely Erased, Webopedia, https://www.webopedia.com/DidYouKnow/Hardware_Software/Erasing_Deleted_Files.asp.

[28] Allen Smith, AI: Discriminatory Data In, Discrimination Out, Society for Human Resource Management (Dec. 12, 2019), https://www.shrm.org/ResourcesAndTools/legal-and-compliance/employment-law/Pages/artificial-intelligence-discriminatory-data.aspx.

Up ↑

Discover more from University of Cincinnati Law Review Blog

Subscribe now to keep reading and get access to the full archive.

Continue reading

Exit mobile version
Skip to content
%%footer%%