The Department of Labor (DOL) has published guidance on using artificial intelligence (AI) tools in employment. The guidance builds on the DOL’s May 2024 AI guidance and fulfills the agency’s obligations under President Biden’s October 2023 executive order on AI.
The DOL disclaimer clearly states that the AI Guidance is not binding and does not supersede, modify, or direct the interpretation of any statute, regulation, or policy. However, it is the DOL’s most comprehensive AI publication. Following the guidance will help employers use AI without violating existing equal employment opportunity and other laws.
The DOL AI Guidance expands on the eight AI principles contained in its May Guidance by providing best practices employers can follow to implement the principles.
Employers should regularly integrate worker input to center worker empowerment. They should ensure that workers and their representatives, especially those from underserved communities, are informed of and have genuine input in the design, development, testing, training, use, and oversight of AI systems in the workplace. By incorporating workers into the process, from design to use, employers can balance the benefits of AI with worker protection and strive to use AI to improve workers’ job quality and enable business success.
To develop AI ethically, employers should develop a strong foundation of ethical standards, guidelines, and an internal review process to help ensure AI and automated systems…meet safety, security, and trustworthy standards for their customers, customers’ workers, and the public. To do this, employers should:
Conduct impact assessments and independent audits of the AI programs and publish the results.
Assess the risks of algorithmic discrimination.
Document negative impacts on workers’ job quality and well-being.
Monitor AI programs continuously and prioritize human oversight over the tools and employment decisions that involve those tools.
Ensure that any jobs created to review and analyze AI comply with domestic and international labor standards.
To establish sufficient AI governance and appropriate human oversight of AI tools, employers should:
Establish empowered governance structures to incorporate workers' input into the decision-making process and continually review and evaluate worker-impacting AI systems.
Offer various employees appropriate training on AI systems, including processes for raising concerns.
Not rely only on AI and automated systems or the information collected to make significant employment decisions.
Identify and document the types of significant employment decisions informed by AI systems, including procedures for human consideration and remedies for decisions that adversely impact employees.
To ensure transparency in AI use, employers should:
Provide workers and representatives with advance notice and appropriate disclosure that AI systems are in use. This information should be clear and accessible, and workers should be conspicuously notified of what data will be collected and stored about them and for what purpose.
Allow employees to view, dispute, and submit corrections for their individually identifiable data without fear of retaliation.
DOL believes this transparency will foster greater trust and job security, prepare workers to effectively use AI, and open channels for workers to provide input to improve the technology or correct errors.
To ensure AI tools do not interfere with employee labor organizing, cause reductions in employee wages, or put employee health and safety at risk, employers should:
Not use AI tools to reduce wages, break time, or benefits.
Audit AI systems for disparate or adverse impacts on individuals with protected characteristics to comply with anti-discrimination requirements, including offering reasonable accommodations when requested.
To use AI to enable workers, employers should:
Create AI pilot programs for employees to use and test tools before conducting large-scale rollouts. This will ensure that the tools assist and complement workers and improve job quality.
Not use AI tools to engage in invasive monitoring of employees, especially when assessing worker performance.
Consider balancing enhanced productivity through AI tools while benefiting workers, such as through increased wages, improved benefits, increased training, fair compensation for collecting and using worker data, or reduced working hours without loss of pay.
To support workers impacted by AI, employers should train employees on AI systems to upskill workers instead of replacing them. Work to preserve jobs for those at risk of displacement due to AI by offering training, education, and professional development opportunities for workers to learn how to use and work with AI systems.
To ensure the responsible use of worker data, employers should:
Develop safeguards for protecting employee data from internal and external threats, emphasizing mitigating worker privacy risks.
Ensure that AI tools have safeguards for securing and protecting data.
Avoid collecting unnecessary data.
Not share data outside of the business.
DOL stresses that employers should follow these eight principles throughout the AI lifecycle. The DOL’s AI Guidance states that the eight principles and best practices are not intended to be an exhaustive list and are not binding. However, the document provides an integral guiding framework for employers to follow to refine how best to use AI in employment decisions.
Employers considering implementing AI systems and procedures should examine the DOL AI Guidance to ensure their systems and procedures track the purposes and policies outlined there so that they keep track of how the government is looking at these issues. Employers should examine the requirements of other federal agencies like the EEOC and the OFCCP, if applicable, and state laws to ensure their systems anticipate all appropriate legal requirements.
Comments