The first agency to formally take notice was the EEOC, which launched a work on AI and Algorithmic Fairness in Hiring Decisions in October 2021 and held a series of listening sessions in 2022 to learn more. In May, the EEOC provided more specific guidance on the use of algorithmic decision-making software and its potential violations of the Americans with Disabilities Act, in a separate support document for employers, including these Without safeguards, the system “risks violating existing laws,” he said. civil rights law. ”
The White House has taken a unique approach, releasing an “AI Bill of Rights Blueprint,” stating that “algorithms used to make employment and credit decisions should reflect and reproduce existing undesirable inequalities, It has been found to embed new and harmful prejudices and discrimination.” On May 4, the White House announced unique commitments from some of the top leaders in the AI space (Anthropic, Google, Hugging Face, Microsoft, NVIDIA, OpenAI, Stability AI) to publicize their AI systems. announced that it will evaluate it to determine its compatibility with AI. AI Bill of Rights.
A joint statement by the FTC, Department of Justice, Consumer Financial Protection Office, and EEOC on April 25 made even stronger language, reiterating its commitment to enforce existing discrimination and prejudice laws. The agency outlined a potential problem with the automation system as follows:
From the Archives (August 2020): Most white people don’t believe racism exists in the workplace, but nearly half of black employees don’t
AI in recruitment is not well regulated
Laws regulating AI are sparse. Of course, there are equal opportunity and anti-discrimination laws that apply to AI-based employment practices. Otherwise, there is no specific federal law regulating the use of AI in the workplace and no requirement for employers to disclose their use of technology.
For now, local governments and states need to shape the new regulatory environment. Two states have passed laws related to consent in video interviews. Starting in January 2020, Illinois enacted a law that requires employers to notify and obtain consent from applicants for the use of AI to analyze video interviews. Since 2020, Maryland has banned employers from using facial recognition service technology on prospective hires unless the applicant signs a waiver agreement.
So far, only New York City has passed legislation specifically to address bias in AI recruitment tools in the United States. The law requires bias audits on automated employment decision tools. It remains unclear how the law will be enforced, as companies lack guidance on how to select reliable third-party auditors. The city’s Department of Consumer and Worker Protection will begin enforcing the law on July 5.
Additional legislation may be enacted. Washington, DC is considering legislation to hold employers accountable for preventing bias in automated decision-making algorithms. California introduced two bills this year aimed at regulating AI in employment. And in late December, New Jersey introduced a bill regulating the use of AI in hiring decisions to minimize discrimination.
SHRM’s Dickens said that at the state and local level, “they’re trying to see if this is something that needs to be regulated, and most importantly, don’t jump into overregulation at the expense of innovation. I think it is,” he said. . ”
AI innovation is moving so rapidly that future legislation is likely to include “flexible and agile” language that takes into account the unknowns, Dickens said.
Also: What skill sets will workers need in the age of AI?
how companies respond
Saira Jesani, deputy secretary general of the Data & Trust Alliance, a nonprofit consortium that guides the responsible application of AI, said HR is “a risk factor for AI,” especially as more companies use AI in their hiring. high applicability”. We build our own tools and buy them.
“Anyone who says AI is bias-free, I don’t think that’s true at this point,” Jesani said. “I say this because I think we are unbiased, and we cannot expect AI to be unbiased.”
But what companies can do is reduce bias and properly vette the AI companies they use, says Jethany, who leads nonprofit initiatives such as developing algorithmic bias safeguards for the workforce. says Mr. These safeguards are used to guide how companies evaluate AI vendors.
He stresses that vendors must demonstrate that their systems can “detect, mitigate and monitor” bias, in case employer data is likely not to be completely bias-free.
“or [employer] The data basically helps train a model on what the output will look like,” Jesani said, emphasizing that companies should look for vendors who take bias seriously into their designs. The data give no clue what the bias is. ”
Read more: Will AI cause mass unemployment? What history tells us about technology and jobs
So will HR robots take over?
AI is evolving rapidly. It’s too fast for this article to keep up with. But despite concerns about the potential for AI to cause stigma and discrimination in the workplace, it’s clear that companies that can afford to adopt it won’t stop using it.
For SHRM’s Alonso, public awareness of AI is a top priority. Regarding the concerns that dominate the debate about AI’s place in employment and beyond, he said:
“There is a fear-mongering argument that ‘AI shouldn’t be introduced’ and ‘AI will eventually learn the biases that exist among developers and then we will start to introduce them’.” There is also the fear-mongering argument, Which is it? [bias] And does it make things more effective in terms of inheriting what we humans have developed and believed? Or is it the fear that AI will eventually take over the whole world?”
Alonso added, “By the time you answer or judge which fear-mongers and fears you are most afraid of, AI will have long since passed us.” rice field.
Learn more about NerdWallet
Anna Herhoski writes for NerdWallet. Email: anna@nerdwallet.com. Twitter: @AnnaHelhoski.
-Anna Herhosky
This content was created by MarketWatch, operated by Dow Jones & Company. MarketWatch is published independently of The Dow Jones Newswire and The Wall Street Journal.
(Closed) Dow Jones Correspondence
06/24/23 1509ET
Copyright (c) 2023 Dow Jones & Company, Inc.
