We’ve reached another milestone in the artificial intelligence revolution: The federal agency charged with enforcing anti-bias laws just recorded its first-ever settlement in a case involving AI discrimination in the workplace.
The Equal Employment Opportunity Commission’s (EEOC) Aug. 9 legal filing in a New York federal court revealed that a tutoring company agreed to pay $365,000 to resolve charges that its AI-powered hiring selection tool automatically rejected women applicants over 55 and men over 60.
An applicant who was rejected from a position at iTutorGroup thought something was fishy when they allegedly submitted their same resume again, but this time included a younger birthdate and secured an interview. They took their complaint to the EEOC, which filed a lawsuit against the employer on behalf of more than 200 applicants, alleging age and gender discrimination. The lawsuit claimed the company illegally screened out women applicants over 55 and men over 60.
iTutor denied the allegations and continues to deny any wrongdoing, despite entering into a voluntary settlement with the EEOC last week. Besides paying $365,000 to a group of more than 200 rejected applicants, iTutor agreed to adopt antidiscrimination policies and conduct employee trainings to ensure compliance with equal employment opportunity laws. The company also must consider anew all the applicants who were purportedly rejected because of their age.
There are two reasons why this settlement is significant in nature. This is a first-of-its-kind settlement. The EEOC has launched a broader initiative to ensure AI workplace tools comply with antidiscrimination laws, and this settlement is a groundbreaking achievement for the agency as it begins this new push. It is certainly not the last one we’ll see over the coming months and years.
We expect to see more legal actions and more settlements because the use of AI in employment settings is exploding. Approximately 79 percent to 85 percent of employers now use some form of AI in recruiting and hiring, and that number will surely increase. Given this exponential rise, employers are bound to have questions about compliance best practices.
10 Pointers to Ensure Compliance
In order to ensure you don’t go down the same path as the company that recently settled its EEOC discrimination claim, here are 10 points you should consider adopting.
- Conduct Diverse Testing: Before fully implementing any AI tools in the HR arena, you should rigorously test them, using diverse data sets. Such a practice ensures that the software won’t inadvertently discriminate against certain demographic groups.
- Regularly Review Your AI-Powered HR Tools: You should continue to periodically review AI tools to make sure things stay compliant. Ensuring that no inherent biases exist in the software is a crucial step in upholding your company’s commitment to diversity. As the EEOC has clearly stated, you can’t pass the buck and blame your software vendor if their AI product ends up committing discriminatory or biased acts with your applicants or employees.
- Conduct Bias Audits: New York City recently became the first jurisdiction to require employers using AI in the employment context to conduct AI bias audits, and it won’t be the last. Even if your organization isn’t subject to the NYC law (or any of the soon-to-be-adopted laws to follow), conducting an AI bias audit (with the help of your legal counsel) could be an invaluable tool in rooting out unintentional discrimination at your workplace.
- Train Your HR Teams: Your HR department should get a crash course on the use of AI in human capital management so they can be your front line when it comes to ensuring fairness. Make sure they have the knowledge and skills to utilize whatever AI tools you adopt without inadvertently perpetuating biases. Besides your regular antidiscrimination training, you should ensure they receive support related to the application and interpretation of AI in all HR functions.
- Establish Clear Workplace Policies: A critical step is developing a thorough workplace AI policy. An explicit and well-communicated policy can act as a foundation for fair HR practices.
- Keep Open Communication: You should foster an environment where applicants and employees are aware of the use of AI in the HR environment, and they feel comfortable voicing concerns about perceived biases. A guidance document issued by the EEOC highlights how an applicant’s or employee’s knowledge of the use of AI in the disability law context could create a pathway to ensure that you provide necessary, reasonable accommodations.
- Don’t Eliminate All the Humans: Your HR professionals should play a vital role in workplace decision-making. We’re being somewhat facetious by even suggesting you can remove all human interaction and replace them with robots. But to the extent that you are incorporating AI technology to supplement and support your HR efforts, you need to make sure you retain a healthy dose of human judgment in your workplace decision-making.
- Incorporate Feedback Loops: No matter how advanced your predictive analytics, it’s hard to predict the real-world dynamics that can arise once you deploy your AI systems. Try to encourage feedback from internal stakeholders – and external candidates and other third parties – regarding their experiences with your AI-driven processes. This will allow you to identify and rectify potential biases or other issues that might arise.
- Seek Out Expertise: Given the complexities of AI and its intersection with workplace law, you should partner with legal counsel who understands the many issues that need to be considered – data privacy, confidentiality, trade secrets, bias audits, copyright law, labor law, and overall best practices, just to name a few.
- Stay Up to Speed: The world is changing at a frenetic pace, especially when it comes to the intersection of AI and human capital management.
Raeann Burgo is an attorney with Fisher Phillips in Pittsburgh. Wendy Hughes is an attorney with Fisher Phillips in Philadelphia. © 2023. All rights reserved. Reprinted with permission.
