Employers using or considering using artificial intelligence AIs that assist workplace operations (AIs) have been reminded again by the federal government that their actions will be subject to rigorous scrutiny by the EEOC for possible employment discrimination violations. A federal agency released a technical assistance document Thursday to help employers deploying AI to assist in hiring and employment-related activities to uncover potential Title VII violations of long-standing legal principles. applied to today’s evolving environment. What are the 5 things you should know about this latest development?
If you want to learn more and tackle more on this subject, sign up for our next bi-weekly episode. Fisher-Phillips AI Workplace Forum on May 24 Discuss this issue further and provide an interactive space to ask questions and participate in the conversation.
1. EEOC Confirms Employer Use of AI May Violate Workplace Laws
The EEOC started by identifying a clear position in the technical assistance document. Improper application of AI may violate Title 7 of the Federal Antidiscrimination Act when used for recruitment, hiring, retention, promotion, transfer, performance monitoring, and demotion. , or fired. The EEOC has outlined four instances of the use of AI during the hiring process and one instance during an employment relationship could result in a Title VII violation.
- resume scanner Prioritize your application with specific keywords.
- “Virtual Assistant” or “Chatbot” Ask job seekers about their qualifications and reject candidates who do not meet predefined requirements.
- video interview software Evaluate candidates based on their facial expressions and speech patterns.
- software testing It provides a ‘job fit’ score on an applicant’s or employee’s personality, aptitude, cognitive skills, or perceived ‘cultural fit’ based on their performance on games and more traditional tests.and
- employee monitoring software Evaluate employees based on keystrokes and other factors.
The agency did not say these were the only workplace-related AI techniques that could come under fire. Nor does it say that such tools are inherently inappropriate or illegal. However, it said that existing government agency regulations (Uniform Guidelines for Employee Selection Procedures) that have existed for more than 40 years can also be applied in situations where employers use AI-powered selection procedures in hiring settings. rice field.
The agency said this was a situation of “heterogeneous influence”, meaning that employers did not intend to discriminate against anyone, but rather put in place some form of neutral process that ultimately resulted in certain protected classes of workers said this is particularly true when it has a statistically significant negative impact on
2. The 4/5 rule can be applied to AI selection
The EEOC noted that employers could use the “four-fifths” rule as a general guideline for determining whether their AI selection process violated different impact criteria (and that the use of mathematics I apologize in advance for the urgency). This test compares the selection rate of a particular group to the most “successful” selection rate to see if the selection process affects particular groups differently. Less than four-fifths of the selectivity can be affected in many ways. Here’s an example provided by the EEOC if that seems confusing.
Suppose your company uses algorithms to score personality tests to determine which applicants make it through the job selection process.
- 80 white applicants and 40 black applicants take a personality test.
- 48 of the Caucasian applicants (representing 60%) advance to the next round.
- Twelve of black applicants advance to the next round (equivalent to 30%).
- So the ratio of the two ratios is 30/60 (or 50%).
- 30/60 (or 50%) is lower than 4/5 (or 80%), so according to the 4/5 rule, black applicants have significantly different selectivity than white applicants . can Evidence of discrimination against black applicants.
Note, however, that the EEOC states that this kind of analysis is just a rule of thumb. This is the basic method for drawing initial inferences about the selection process. According to the EEOC, if you do eventually find a problematic figure, it should prompt you to obtain additional information about the procedure in question, and it doesn’t necessarily indicate a conclusive Title VII violation. Similarly, just because a number clears the 4 out of 5 hurdle does not mean that a particular selection procedure is definitely legal under Title VII. Allegations of discrimination can still be challenged by authorities or plaintiffs.
3. EEOC encourages proactive self-auditing
In a statement accompanying the release of the technical assistance document, EEOC Chairman Charlotte Burroughs said employers should test all employment-related AI tools early and often to ensure they do not cause legal harm. said there was a need. This doesn’t just mean he uses the 4/5 rule, it also means he uses a thorough audit process that includes a variety of potential inspection methods for all AI capabilities. “Employers are encouraged to conduct ongoing self-analysis to determine whether they are using technology in a way that may cause discrimination,” she said.
But the EEOC did not mention it. This is a reminder that self-audits should be undertaken with the help of legal counsel. Experienced attorneys not only guide you on the best methodologies to use and assist you in interpreting audit results, but also use attorneys to mask and identify acts of attorney-client privilege. result from detection. This is especially useful for identifying changes needed to improve processes and minimize unintended impacts.
4. Facing Problems Caused by AI Vendors
The agency also explicitly pointed out that liability cannot be avoided by using a third party to introduce AI techniques and then blaming the resulting discriminatory consequences on the third party. It states that even if the decision-making tool was developed by an outside vendor, it could still be held liable if the AI procedure discriminated on grounds prohibited by Title VII. increase.
“Furthermore, an employer may be liable for the conduct of its agents, including entities such as software vendors, if the employer authorizes the agent to act on the employer’s behalf,” the EEOC said. said. This may include situations where an agent is dependent on the outcome of a selection process conducted on your behalf.
The EEOC specifically asks vendors considering developing or maintaining algorithmic decision-making tools whether steps have been taken to assess whether the tools may have various adverse effects. recommend to. It also asks vendors whether they rely on the four-fifths rule of thumb or criteria such as statistical significance, which courts often use when investigating employer behavior for potential Chapter 7 violations. It is recommended to ask whether the
5. EEOC guidance is part of a larger trend
This technical assistance document is part of a broader trend we are seeing from federal agencies growing concerned about the potential for AI to lead to employment law violations. In fact, just last month, EEOC Chairman Burroughs worked with leaders from the Department of Justice, the Federal Trade Commission, and the Consumer Financial Protection Agency to address potential employment-related biases that can arise from the use of AI and algorithms. announced that it would be investigated. at work.
And within the past year, the EEOC has worked with the Justice Department to warn that relying on AI to make staffing decisions could unintentionally lead to discriminatory employment practices, including prejudice against people with disabilities. , followed by the White House’s release of a “blueprint.” The AI Bill of Rights aims to protect civil rights in building, deploying, and governing automated systems.
None of these guidance documents create new legal standards and cannot be relied upon as enforceable as statutes and regulations, but they are important and indicate where government agencies are focusing. It may indicate that you are conducting enforcement activities and may be cited by government agencies and agencies. Introduce plaintiffs’ attorneys as best practices for employers to follow. And states are taking action as well, with New York City’s law set to take effect in July and California’s new bill being presented to the governor. As such, we need to take this guidance seriously, adjust our hiring practices where necessary, and keep up with the pace of change that is rapidly unfolding before us.
Want to know more?
This latest guidance, as well as a list of AI-related legal obligations that employers are likely to face on an almost daily basis, will be the subject of the next Fisher-Phillips AI Workplace Forum on Wednesday, May 24th. This free event will discuss these issues in more detail and provide an interactive space to ask questions and participate in the conversation. You can register for the event here.
If you have questions about how best to maximize the value of AI in the workplace while mitigating legal, ethical, and reputational risks, please contact a Fisher-Phillips attorney, the author of this insight, or the Artificial Intelligence Practice Please contact a group attorney. We will continue to monitor future developments and provide updates on this and other workplace law issues. So please make sure you are subscribed to the Fischer-Philips Insight system to collect the latest information.
