Why AI+HI Is Essential to Compliance

HR must always include human intelligence (HI) and oversight of artificial intelligence (AI) in decision-making in hiring and firing, according to Kelly Dobbs Bunting, an attorney with Greenberg Traurig in Philadelphia. Even though there are currently few AI laws in the U.S., more will be coming. Bunting explained that companies can ensure compliance by meeting the strictest AI standards, which will be in Colorado’s AI law, taking effect Feb. 1, 2026. However, some provisions of that law might be scaled back.

Right to Appeal Sparks Concern

Speaking June 24 at the SHRM Annual Conference & Expo 2024 (SHRM24), Bunting noted that under the Colorado law, a person subject to an adverse employment action as a result of AI has the right to appeal. She said she’s not certain how that will play out, but anyone who doesn’t get a job might have the right to go to the company and appeal. Any bad or even average performance review in theory might be appealed. Not receiving a raise or promotion might be appealed as well.

This right to appeal is “concerning,” Bunting said, adding that the law is still being discussed and might be modified. Colorado Gov. Jared Polis has pledged to help bring about changes to the law, and business groups are likely to say the right-to-appeal provision will be a “nightmare,” Bunting said.

Other Requirements

Other provisions in Colorado’s AI law include a requirement that any “deployer” of AI—that is, an individual or entity doing business in the state—report to the Colorado attorney general if it knows it has caused harm due to AI.

Noncompliance with the law may be a deceptive trade practice, which only the attorney general will be able to prosecute. An employee may not bring a private right of action, Bunting noted.

Employers engaged in “high-risk” practices, such as rejecting a job applicant or firing someone based on information derived from AI, must take reasonable care to avoid algorithmic discrimination in the state.

Policies related to AI must be in place, and there must be a review of AI tools every year—a review most likely conducted by a third party, she said. Employers also must give job applicants and employees notice that AI systems are being used. Small employers with fewer than 50 employees are exempt from some of the law’s provisions.

New York City has an AI bill as well, but there hasn’t been much litigation under it, and some experts consider it so far to be a bust, Bunting said.

But more state AI laws are on the way, with ones expected in California and Washington state soon, she said. Bills in both states would require:

• Prohibiting AI discrimination.
• Annual audits.
• Notice that software is being used.

If the Washington bill passes, it would take effect Jan. 1, 2025, Bunting said.

Best Practices

 Bunting recommended several best practices for employers using AI:

• Conduct regular audits.
• Review to ensure vendor compliance.
• Provide notice.
• Get consent from applicants and employees to use AI.
• Provide alternatives if they don’t want to be subject to it.
• Create and comply with HR policies.
• Train HR and managers.

“Always have humans overseeing decisions,” Bunting emphasized. Employers can’t just hand off everything to AI software.

HR professionals almost have “to become lawyers” and think about what they would show the U.S. Equal Employment Opportunity Commission or state agencies to show they tried in a good-faith manner to prevent a bad outcome, Bunting said.