On 16 October 2024, the Division of Labor (DOL) printed a complete steering concerning the usage of synthetic intelligence (AI) instruments in employment. The steering, entitled “Artificial Intelligence and Employee Effectively-being: Ideas and Greatest Practices for Builders and Employers”1 (the DOL AI Steerage), builds on the DOL’s Could 2024 AI steering (the Could Steerage) and fulfills the company’s obligations underneath President Biden’s October 2023 govt order on AI.2 The DOL AI Steerage additionally follows the company’s endorsement of the Partnership on Employment & Accessible Know-how (PEAT)’s AI & Inclusive Hiring Framework,3 in addition to publications by DOL subagencies just like the Equal Employment Alternative Fee (EEOC)4 and the Workplace of Federal Contract Compliance Applications (OFCCP).5
The DOL makes clear in its disclaimer that the DOL AI Steerage will not be binding and doesn’t supersede, modify, or direct an interpretation of any statute, regulation, or coverage. Nonetheless, the publication is the DOL’s most complete AI publication up to now, and following the steering will assist employers use AI with out operating afoul of present equal employment alternative and different legal guidelines.
Ideas and Greatest Practices
The DOL AI Steerage expands on the eight AI ideas (Ideas) contained within the Could Steerage by offering greatest practices (Greatest Practices) that employers6 can observe to implement these Ideas.
Centering Employee Empowerment
First, to “middle employee empowerment” by guaranteeing that “[w]orkers and their representatives, particularly these from underserved communities, [are] knowledgeable of and have real enter within the design, growth, testing, coaching, use, and oversight of AI techniques within the office,” employers ought to often combine enter from employees. By incorporating the employee into the method, from design to make use of, employers can stability the advantages of AI with employee safety and try to make use of AI to enhance employees’ job high quality and allow companies success.
Ethically Creating AI
Second, to ethically develop AI, employers ought to develop a powerful basis consisting of moral requirements, pointers, and an inner assessment course of to assist “guarantee AI and automatic techniques…meet security, safety, and reliable requirements for his or her prospects, prospects’ employees, and the general public.” To do that, employers ought to do the next:
- Perform impression assessments and impartial audits of the AI applications and publish the outcomes.
- Assess the dangers of algorithmic discrimination.
- Doc destructive impacts on employees’ job high quality and well-being.
- Monitor the AI applications on an ongoing foundation and prioritize human oversight over the instruments and employment choices that contain these instruments.
- Be sure that any jobs created to assessment and analyze AI adjust to home and worldwide labor requirements.
Establishing AI Governance and Human Oversight
Third, to ascertain adequate AI governance and applicable human oversight of AI instruments, employers ought to do the next:
- Set up empowered governance constructions to include enter from employees within the decision-making course of to repeatedly assessment and consider worker-impacting AI techniques.
- Supply applicable coaching on AI techniques to a broad vary of workers, together with processes for elevating considerations.
- Not rely solely on AI and automatic techniques, or the data collected by way of them, to make vital employment choices.
- Determine and doc the forms of vital employment choices knowledgeable by AI techniques, together with procedures for human consideration and treatments for choices that adversely impression workers.
Making certain Transparency in AI Use
Fourth, to make sure transparency in AI use, employers ought to do the next:
- Present employees and representatives advance discover and applicable disclosure that AI techniques are in use. This data needs to be clear and accessible, conspicuously notify employees what knowledge might be collected and saved about them, and what that knowledge might be used for.
- Enable workers to view, dispute, and submit corrections for his or her individually identifiable knowledge with out concern of retaliation.
Based on the DOL, this transparency will “foster higher belief and job safety, put together employees to successfully use AI, and open channels for employees to offer enter to enhance the expertise or appropriate errors.”
Defending Labor and Employment Rights
Fifth, to make sure AI instruments don’t intrude with workers’ labor organizing, trigger reductions in workers’ wages, or put workers’ well being and security in danger, employers ought to do the next:
- Not use AI instruments to cut back wages, break time, or advantages.
- Audit AI techniques for disparate or hostile impacts on people with protected traits to adjust to anti-discrimination necessities, together with providing cheap lodging when requested.
Utilizing AI to Allow Staff
Sixth, to make use of AI to allow employees, employers ought to do the next:
- Create AI pilot applications for workers to make use of and check instruments earlier than conducting large-scale rollouts to make sure the instruments are aiding and complementing employees and enhancing job high quality.
- Not use AI instruments to have interaction in invasive monitoring of workers, particularly when assessing employee efficiency.
- Take into account how you can stability enhanced productiveness by way of the usage of AI instruments whereas benefiting employees, reminiscent of by way of “elevated wages, improved advantages, elevated coaching, truthful compensation for the gathering and use of employee knowledge or diminished working hours with out lack of pay.”
Supporting Staff Impacted by AI
Seventh, to assist employees impacted by AI, employers ought to do the next:
- Practice workers on AI techniques to upskill employees as a substitute of changing them.
- Work to protect jobs for these vulnerable to displacement attributable to AI by providing coaching, training, {and professional} growth alternatives for employees to learn to use and work with AI techniques.
Making certain Accountable Use of Employee Knowledge
Eighth, to make sure accountable use of employee knowledge, employers ought to do the next:
- Develop safeguards for shielding worker knowledge from inner and exterior threats, with an emphasis on mitigating privateness dangers for employees.
- Be sure that AI instruments have “safeguards for securing and defending knowledge.”
- Keep away from accumulating pointless knowledge.
- Not share knowledge outdoors of the enterprise.
Takeaways
The DOL stresses that employers ought to make the most of every of those eight Ideas “throughout the entire lifecycle of AI – from design to growth, testing, coaching, deployment and use, oversight, and auditing.” Additional, the DOL clarified in its DOL AI Steerage that the eight Ideas and the Greatest Practices it outlined aren’t supposed to be an “exhaustive listing” and, as famous above, aren’t binding. Nonetheless, the doc supplies an integral “guiding framework” employers can observe as they refine how greatest to make use of AI in employment choices.
Suggestions
Employers which are implementing or contemplating implementing AI techniques and procedures ought to look at the DOL AI Steerage to make sure their techniques and procedures monitor the needs and insurance policies outlined within the Ideas and Greatest Practices. Employers additionally ought to proceed inspecting necessities of different federal companies—such because the EEOC and the OFCCP, if relevant—in addition to state legal guidelines to make sure their techniques meet all applicable authorized necessities.