U.S. Department of Labor Issues Additional Guidance Regarding AI and Workers’ Well-Being
Rob Thomas, Of Counsel
The Rocky Mountain Employer recently discussed the U.S. Department of Labor’s (“DOL’s”) recent guidelines focused on preventing disability discrimination in the hiring process resulting from the use of artificial intelligence (“AI”),[1] and has previously discussed Colorado’s own legislation intended to curb unintentional discrimination resulting from the use of AI in employment decisions.[2] Now, the DOL has issued additional practical guidance for employers and AI developers centered on the ethical use of AI with an eye towards improving workers’ lives and augmenting existing work opportunities, rather than displacing workers or diminishing their role in the marketplace.
AI and Worker Well-Being – Purpose and Authority
The DOL’s recent guidance—Artificial Intelligence and Worker-Well Being: Principles and Best Practices for Developers and Employers (the “Best Practices”)—is responsive to President Biden’s October 23, 2023, Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence and is intended to further that Executive Order’s goals in maximizing the potential benefits of AI in the workplace while mitigating potential harms to the workforce.[3] The DOL drafted the Best Practices with input from workers, labor unions, researchers, academics, employers, developers, and others in order to facilitate existing efforts (particularly between unions and employers) to set reasonable “guardrails” to protect workers from over-encroachment of AI in the workplace. While the Best Practices do not have the force of law and do not otherwise supersede any existing laws or regulations, the Best Practices are nonetheless intended to provide a framework for employers looking to further weave AI into their operations while protecting existing workers’ rights.
The Best Practices, and the “North Star” of Centering Worker Empowerment
The Best Practices set forth eight overall goals, with “Centering Worker Empowerment” listed as the “North Star” goal of the Best Practices:
· Centering Worker Empowerment – First and foremost, the Best Practices emphasize the importance of worker input throughout the AI development cycle—from design and testing to deployment, use, and oversight—and particularly from historically underserved communities.
· Ethically Developing AI – This Best Practice emphasizes the importance of clear ethical standards and review processes in the development and implementation of AI in the workplace in order to prevent harms to workers’ civil rights, mitigate safety risks, meet performance requirements, and avoid disparate impact discrimination. Any AI development and implementation should be subject to impact assessment reviews and independent audits of any potential harms caused by the AI systems at issue.
· Establishing AI Governance and Human Oversight – This Best Practice emphasizes the importance of human oversight and governance over any worker-impacting AI systems, including robust training on AI, corporate governance structures, and avoiding overreliance on AI and other automated systems when making significant employment decisions.
· Ensuring Transparency in AI Use – This Best Practice strongly encourages employers to disclose in advance to workers and labor unions any worker-impacting AI systems before they are launched or implemented. This Best Practice includes disclosing any data collected or stored about workers and for what purposes such data may be used.
· Protecting Labor and Employment Rights – This Best Practice warns against any use or implementation of AI systems which might negatively impact or chill workers’ rights to organize or engage in other protected activities under the law (such as discussing wages and working conditions), AI usage that might negatively impact worker health and safety (such as automated systems which reduce reporting of injuries or disincentivize taking rest breaks or working at a reasonable pace), AI usage resulting in disparate impact discrimination (including disability), or AI usage resulting in wage and hour violations.
· Using AI to Enable Workers – This Best Practice emphasizes that any gains in efficiency or revenue resulting from the implementation of AI should not come at the expense of workers, and should be used to support productivity, performance, and worker well-being.
· Supporting Workers Impacted by AI – This Best Practice suggests ways for employers to minimize the impact to workers displaced by AI, including training opportunities related to AI to prevent displacement before it occurs, or retraining and reallocating workers to other jobs within an organization if they have been displaced.
· Ensuring Responsible Use of Worker Data – Last, this Best Practice emphasizes the protection, non-disclosure, and limited use of worker data collected for business purposes, with appropriate safeguards and other breach-prevention measures.
Employer Considerations
Again, while these Best Practices do not have the force of law, they clearly signal that the DOL currently takes the impact of AI on worker well-being seriously, and that any employer abuses of AI in the workplace which diminish worker rights or violate federal laws will draw the DOL’s attention. Campbell Litigation, P.C. is available for consultation on matters related to these DOL guidelines and other employment law concerns.
[1]See https://www.rockymountainemployersblog.com/blog/2024/10/10/us-department-of-labor-issues-framework-to-guide-employers-implementing-artificial-intelligence-technology-in-the-hiring-process.
[2]See https://www.rockymountainemployersblog.com/blog/2024/5/30/colorado-pioneers-law-aimed-at-addressing-discrimination-in-artificial-intelligence-systems.
[3]See https://www.dol.gov/sites/dolgov/files/general/ai/AI-Principles-Best-Practices.pdf for a full copy of the DOL’s recent guidance; see also https://www.whitehouse.gov/briefing-room/presidential-actions/2023/10/30/executive-order-on-the-safe-secure-and-trustworthy-development-and-use-of-artificial-intelligence for President Biden’s Executive Order.