Federal Agencies Release Joint Statement on AI

10 federal agencies released a joint statement on April 4 expressing their commitment to protecting the public from unlawful bias in automated systems. Automated system is broad language that is described as “software and algorithmic processes, including AI, that are used to automate workflows and help people complete tasks or make decisions.” Automated systems and AI have and will continue to become more prominent in our society, as these systems can provide increasingly sophisticated solutions to complex problems, revolutionizing industries and enhancing efficiency and productivity.

Many organizations already use these systems to help with different types of decisions related to individual rights and opportunities such as housing, credit opportunities, and hiring. One of the major concerns regarding automated systems is that they may result in discriminatory decision making. Automated systems are reliant on the people and data so if the data is skewed, or the developer doesn’t account for the right context and situation, the use of an automated system could lead to unlawful discrimination. Another concern is that many automated systems are considered “black boxes” meaning that the inner workers or the decision-making process used by the systems are not transparent or understandable to users. In other words, a user can input data into the system and then receive an output or decision, but they do not know how the system arrived at the output. Without understanding how the system behaves, its hard to identify whether there was fairness or potential bias involved in the decision-making process.

In the joint statement, the federal agencies express their understanding of the seriousness around these automated systems and their responsibilities to ensure that ensuring that these systems are developed and used in accordance with federal laws. The joint statement also included examples of the concerns each agency has expressed in the past regarding the harmful uses of automated systems. For example, the EEOC has issues two technical assistance documents regarded automated systems with one pertaining to how the ADA applies to the use of automated systems when making employment-related decisions, and the other pertaining to how using automated systems can lead to disparate impact under Title VII. Another example included the OFCCP’s release of their FAQ explaining how they will analyze AI-based selection procedures that have adverse impact the same as they analyze other selection practices and the update to their compliance review process to require documentation related to automated systems used by federal contractors when making hiring decisions.

Employers who use automated systems should review how federal and state agencies are analyzing these systems to ensure compliance with the law. It’s recommended to review all systems used for employment-related decisions to evaluate the risk of bias from using these systems. As automated systems and AI continue to become a prevalent tool in the workforce, more laws and regulations are expected to emerge to ensure fairness and compliance.