EEOC issues new guidance on employer use of AI, and compliance with the Americans with Disabilities Act.

ABSTRACT: New guidance from the EEOC comes at the right time with over 80 percent of employers using AI in employment decision-making. The guidance highlights common ways discrimination occurs and best practices to avoid the courtroom.

On May 12, 2022, the Equal Employment Opportunity Commission issued a new comprehensive “technical assessment” document to aid employers with navigating existing ADA requirements when using software, algorithms, and artificial intelligence (collectively, “AI”) in employment-related decision making.

This is part of the Justice Department and the EEOC’s joint initiative to combat discrimination against people with disabilities. EEOC Chair Charlotte Burrows, citing data from the U.S. Bureau of Labor Statistics, stated “...over 80 percent of employers use AI in some form in their broader work and their employment decision-making...”

The new guidance highlights potential issues and scenarios that could lead to discrimination claims based on AI use and recommends best practices. Employers may be familiar with some recommendations, as its best practices are already reflected in some state laws, including the 2020 Illinois Artificial Intelligence Video Interview Act.

The EEOC identified three common ways discrimination can occur with AI:

1. An employer does not provide a “reasonable accommodation” that is necessary for a job applicant to be rated fairly and accurately by the algorithm;
2. An employer intentionally, or unintentionally, “screens out” an individual with a disability;
3. An employer adopts an algorithmic decision-making tool for use with its job applicants or employees that violates the ADA’s restrictions on disability-related inquires and medical examinations.

In many cases an employer can still be responsible under the ADA even if the tools are designed or administered by a third-party vendor.

A. Reasonable Accommodations

EEOC recommends proactively telling job seekers that the evaluation process uses AI and whether a candidate will need a reasonable accommodation to complete it. If an applicant were to tell a vendor that a medical condition was making it difficult to take the test, an employer would likely be responsible even if it was unaware the applicant reported a problem to the vendor.

Employers must promptly respond to requests for reasonable accommodations. Once it is shown a disability might make a test more difficult to take, or reduce the accuracy of an assessment, an employer must provide an alternative testing format, or more accurate assessment, unless doing so would involve an “undue hardship.”

B. Screen Outs

Screen out occurs when a disability prevents a job applicant or employee from meeting – or lowering their performance – on selection criterion. The applicant or employee then loses a job opportunity as a result. A screen out becomes unlawful when the individual who is screened out is able to perform the essential functions of the job with a reasonable accommodation.

Does an employer use a chatbot that screens out applicants with a significant gap in employment history? Was the gap caused by a disability? A visual memory assessment could screen out blind candidates able to perform the essential functions of a job that requires a good memory.

Employers can run into problems when they use AI to hire employees who are most similar to their successful employees based on data. For example, AI may discover a candidate’s ability to handle distraction, measured by a test, could be correlated with a successful employee. However, a candidate with PTSD may score poorly on this test. Simple noise-cancelling headphones or a quiet work station could serve as a reasonable accommodation to avoid a screen out of this candidate. 

C. ADA’s Restrictions on Disability-Related Inquires and Medical Examinations

An employer might violate the ADA if it uses AI that poses “disability-related inquires” or seeks information that qualifies as a “medical examination” before giving a candidate offer of employment. Assessments seek “disability-related inquiries” if questions are likely to elicit information about a disability or directly asks whether an applicant or employee is an individual with a disability. An assessment is a “medical examination” if it seeks information about an individual’s physical or mental impairments or health. 

An algorithmic decision-making tool that could be used to identify an applicant’s medical conditions would violate these restrictions if it were administered prior to a conditional offer of employment. Once employment has begun, disability-related inquires may be made and medical examinations may be required only if they are legally justified under the ADA.

D. Best Practices and Third Parties

Vendors advertising their algorithmic decision-making are “bias-free” typically means the programmer has taken steps to prevent discrimination under Title VII, which is discrimination based on race, sex, national origin, color or religion. This does not mean the vendor has taken steps needed to address the problem of disability bias.

Ask questions! Employers can ask the vendors whether the AI was developed with individuals with disabilities in mind. Was the interface designed to be accessible to as many individuals as possible? Can the materials be presented in alternative formats? What steps did the vendor take to determine whether the algorithm disadvantages individuals with disabilities?

Employers should consider entering into agreements with vendors that require them to provide reasonable accommodations on the employer’s behalf.

No matter who develops the AI, important steps to reduce the chance a tool will screen out someone because of a disability includes:

1. Clearly indicating reasonable accommodations, including alternative formats and alternative tests, are available to people with disabilities;
2. Providing clear instructions for requesting reasonable accommodations; and
3. In advance of the assessment, providing all job applicants and employees who are undergoing assessment with as much information about the AI as possible.

At the end of the day, employers should try to ensure decision-making tools measure abilities or qualifications that are truly necessary for the job. This means measuring abilities or qualifications directly when possible. Employers can also train staff to recognize and process requests for reasonable accommodations. Questions concerning an employee or applicant’s disability and his ability to perform the essential functions of a job are often unique, and employers using IT as a screening device may need to take steps beyond what is necessary to address other forms of discrimination.