- Home
- Newsletters Home
- DOL’s AI Hiring Framework
DOL’s AI Hiring Framework
Editor's Note
DOL's AI Hiring Framework
The Department of Labor, along with some great help, has announced a 10 step framework for AI and Inclusive Hiring. Good for the DOL. They are thinking about risks of AI in hiring, getting input from groups representing people who may be harmed, and reminding employers that there are pros and cons in using AI assisted tools.
The 10 steps in the framework are really broad. It's a framework. One of the steps is, "Assess possible positive and negative impacts," which essentially means, "imagine what could possibly go wrong." It's a good and important thing to do. It's also impossible for people who are still learning about AI, how it works (to the extent we can tell), and what the potential problems, risks, and benefits are.
When experts who understand the territory design processes for people who may not be experts, there can be a mismatch between what we need to know to make the framework useful and what most of us know. Along with that, we also don't know what we don't know.
This is the biggest problem with GenAI. It creates stuff that sounds plausible, but it's making stuff up based on probabilities, not our human reality. When you read it, it's easy to think, "Okay. Too many buzzwords, but it seems to say something here." It's hard to know when what it says is completely fictional or just plain wrong. When it comes to AI, not knowing what we don't know is a big deal. I'm already seeing complete nonsense related to employment law issues and terrible advice, which brings me back to the AI and Inclusive Hiring Framework.
The DOL worked with Partnership on Employment & Accessible Technology (PEAT) to develop the framework. The PEAT website resources on the AI and Inclusive Hiring Framework are extremely well done. There are different sections based on who you are and what you are trying to do. Each step is explained and there are guides to help you work that step. There's even a glossary and additional resources broken out by focus area.
As we learn, we create shortcuts in language and ideas so we can cover territory easily and move on to the next thing we're figuring out. It's wonderful when organizations and experts come together to explain complex concepts and how to work with them. It makes learning and compliance easier for all of us. What a great thing.
- Heather Bussing
Companies are increasingly turning to artificial intelligence (AI) to assist with employment related tasks, such as recruiting and hiring. AI tools are useful to increase efficiency, streamline the recruiting process, and eliminate human bias. However, these tools can pose a risk of inadvertent discrimination against job applicants. We recently published an Employer’s Guide to Outsmarting Artificial Intelligence Liability in the Workplace, outlining ways employers can take advantage of AI tools while minimizing legal risks associated with this technology. As an additional resource, the U.S. Department of Labor (DOL) recently promoted a new AI & Inclusive Hiring Framework, which “offers guidance for employers, job seekers, and workers to maximize benefits and better manage risks when obtaining and using AI hiring technology.” DOL promoted the framework in alignment with the Biden Administration’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence.
According to the DOL’s press release, the framework seeks both to “help employers reduce the risks of creating unintentional forms of discrimination and barriers to accessibility as they implement AI hiring technology,” and to “help workers and job seekers navigate the potential benefits and challenges they may face when encountering AI-enabled technologies.”
The AI Framework was published by the Partnership on Employment & Accessible Technology (PEAT), a private company, with funding from the DOL’s Office of Disability Employment Policy (ODEP). PEAT’s stated mission is “to foster collaborations in the technology space that build inclusive workplaces for people with disabilities” with a vision of “a future where new and emerging technologies are accessible to the workforce by design.”
According to the DOL’s press release, the framework is based on the National Institute of Standards and Technology’s AI Risk Management Framework (NIST) and includes NIST’s best practices for inclusive hiring. Notably, the website contains a disclaimer that “PEAT material does not necessarily reflect the views or policies of ODEP or DOL.”
The Framework
The framework consists of 10 focus areas for employers:
- Identify employment and accessibility legal requirements
- Establish staff roles, responsibilities, and training
- Inventory and classify technology
- Develop policies and procedures to work with vendors
- Assess possible positive and negative impacts
- Create a process to provide accommodations for job seekers
- Understand and explain how the AI technology works
- Ensure human oversight
- Manage incidents and appeals regarding AI technology failures.
- Monitor regularly the AI technology’s performance, trustworthiness, and fairness
Within these focus areas are practices, goals, and sample activities that employers can adopt in their AI governance and disability-inclusive hiring initiatives. The website notes that employers do not need to implement “every practice, goal, and activity at once,” but rather, the framework is a “progressive effort and will evolve over time as your organization’s AI operation grows.” Although the focus areas are numbered, employers are encouraged to start with the focus area that makes the most sense to their business, and to skip around at their own pace.
Takeaways for Employers
In terms practical application, the framework mostly builds on prior federal guidance, including the DOL’s AI Principles and the Biden Administration’s Executive Order. However, in a sea of information, the framework can serve as a helpful starting point for employers who wish to take advantage of AI technology while minimizing associated legal risk.
