Available AI tools

The landscape of AI technology is continually growing and covers all portions of the hiring process — recruiting, interviewing, selection, and onboarding. Some companies use automated candidate sourcing technology to search social media profiles to determine which job postings should be advertised to particular candidates. Others use complex algorithms to determine which candidates' resumes best match the requirements of open positions. And some employers use video interview software to analyze facial expressions, body language, and tone to assess whether a candidate exhibits preferred traits.

Federal anti-discrimination law

Although AI tools likely have no intent to unlawfully discriminate, that does not absolve them from liability. This is because the law contemplates both intentional discrimination (disparate treatment) as well as unintentional discrimination (disparate impact). The larger risk for AI lies with disparate impact claims. In such lawsuits, intent is irrelevant. The question is whether a facially neutral policy or practice (e.g., use of an AI tool) has a disparate impact on a particular protected group, such as on one's race, color, national origin, gender, or religion.

The Equal Employment Opportunity Commission, the federal agency in charge of enforcing workplace anti-discrimination laws, has demonstrated an interest in AI and has indicated that such technology is not an excuse for discriminatory impacts.

Discrimination associated with AI tools

The diversity of AI tools means that each type of technology presents unique potential for discrimination. One common thread, however, is the potential for input data to create a discriminatory impact. Many algorithms rely on a set of inputs to understand search parameters. For example, a resume screening tool is often set up by uploading sample resumes of high-performing employees. If those resumes favor a particular race or gender, and the tool is instructed to find comparable resumes, then the technology will likely reinforce the existing homogeneity.

Some examples are less obvious. Sample resumes may include employees from certain zip codes that are home to predominately one race or color. An AI tool may favor those zip codes, disfavoring applicants from other zip codes of different racial composition. Older candidates may be disfavored by an algorithm's preference for ".edu" email addresses. In short, if a workforce is largely comprised of one race or one gender, having the tool rely on past hiring decisions could negatively impact applicants of another race or gender.

Steps to mitigate risk

There are a handful of steps that employers can take to use these technologies and remain compliant with anti-discrimination laws.

First, companies should demand that AI vendors disclose as much as possible about how their products work. Vendors may be reticent to disclose details about proprietary information, but employers will ultimately be responsible for discriminatory impacts. Thus, as part of contract negotiations, a company should consider seeking indemnification from the vendor for discrimination claims.

Second, companies should consider auditing the tool to ensure it does not yield a disparate impact on protected individuals. Along the same lines, companies should be careful in selecting input data. If the inputs reflect a diverse workforce, a properly functioning algorithm should, in theory, replicate that diversity.

Third, employers should stay abreast of developments in the law. This is an emerging field and state legislators have taken notice. Illinois recently passed regulation governing the use of AI in the workplace and other states, including New York, have introduced similar bills.

AI can solve many hiring challenges and help cultivate a more diverse and qualified workforce. But the tools are often only as unbiased as the creators and users of that technology. Careful implementation will ensure AI becomes a discrimination solution — not a landmine.