A stakeholder panel hosted by the U.S. Equal Employment Opportunity Commission (EEOC) in September was dominated by discussions related to smart technologies, such as machine-learning recruiting systems used for hiring.
The meeting was for employers, lawyers and civil rights advocates to discuss topics the agency needs to consider as it drafts its new Strategic Enforcement Plan (SEP).
Witnesses spoke about pregnancy accommodations, caregiver gender discrimination and pay equity as issues to consider for the next SEP, which provides organizations with a roadmap of areas where they can expect to increased monitoring in the future.
But the discussion of AI biases made up the bulk of the five-hour listening session.
“[We] urge future SAs to emphasize how the use of algorithmic technology and artificial intelligence in hiring can replicate and systematize the [stereotypical] decision-making while making that discrimination harder to challenge because of the black box nature of those decision-making processes,” said Emily Martin, vice president for education and workplace justice at National Women’s Law. Center.
Judy Conti, director of government affairs at the National Employment Law Project, said automation tools are becoming more common, but they’re often programmed and trained based on past hiring practices that can replicate illegal bias patterns.
“They’re looking for metrics that are supposed to correlate with job performance, but those metrics may well be built on pre-conceived notions that embody implicit biases and could filter out qualified candidates who don’t fit a certain mold,” he said. she noted.
AI has come under intense scrutiny
In 2019, The Washington Post reported that an algorithmic hiring system developed by Utah-based technology company HireVue evaluated more than one million video job interviews.
Its self-driving interview system asks candidates questions, films their responses and then uses the video to assess candidates for different roles based on an “employability” score, which takes into account their “willingness to learn” and their “personal stability”. HireVue has since stopped its facial analysis.
Meredith Whittaker, co-founder of the AI Now Institute, a research center in New York, said The Washington Post that such algorithmic systems are “pseudoscience” and “license to discriminate against” underrepresented candidates. But HireVue says it uses “world-class bias testing” techniques to prevent hiring discrimination.
In 2018, tech giant Amazon scrapped an AI-based recruiting tool after finding it discriminated against women. The algorithm was based on the number of resumes submitted over the past decade, most of which were from men. Therefore, it was formed to favor men over women.
And in 2022, the EEOC sued three integrated companies providing English tutoring services to students in China under the “iTutorGroup” brand, alleging they programmed their online software to automatically reject more than 200 older applicants.
Eve Hill, a disability rights attorney and partner at Brown Goldstein & Levy in Baltimore, told the EEOC meeting that AI-based screening providers should teach and test the tools on large sets of data documented and diversified on an ongoing basis to ensure results don’t discriminate.
“AI-based discrimination is one of the things that scares me the most right now,” Hill said. “Employers who use them often don’t know how they work.”
AI can be an effective tool when used correctly
Emily Dickens, chief of staff, head of government affairs and general secretary of the Society for Human Resource Management (SHRM), noted that nearly 80% of organizations surveyed in a recent survey were using or planning to use AI for HR purposes over the next five years. years.
She also said leveraging AI-powered devices in HR isn’t so bad. For example, algorithmic systems have transformed the way businesses operate by reducing the time it takes to fill vacancies. And nearly 3 in 5 organizations report that the quality of recruits is higher due to their use of AI.
A 2022 SHRM report showed that AI can help eliminate unconscious bias among HR managers, leading to increased equal employment opportunity when done correctly.
“Now is not the time to impose heavy-handed regulatory restrictions that will set back key HR functions and hamper the ability to create and identify talent pipelines,” Dickens said.
Several witnesses praised the EEOC for its AI and Algorithmic Fairness Initiative, which ensures that AI is used fairly and consistently with federal laws. And some municipalities have implemented regulations to address workplace biases related to AI.
For example, New York City has passed a law that will prohibit companies from using AI and algorithm-based technologies for recruitment, hiring, or promotion without first having these tools audited by a third parties to identify any bias.
“When I look across the country in the last two or three years, almost every state is starting to look at AI and [create] a task force,” said Darrell Gay, partner at the law firm ArentFox Schiff in Washington, D.C. “I urge the EEOC [to create its] own working group in coordination with other agencies to look at this systematically across the country. »