EEOC Relies on AI-Based Recruitment Tools May Distinguish: NPR

AI may be the recruiting tool of the future, but it may come with the legacy effects of discrimination.

With nearly all large employers in the United States now using artificial intelligence and automation in their hiring processes, the agency that enforces federal anti-discrimination laws is considering some urgent questions:

How can you prevent employment discrimination when a machine perpetuates discrimination? What kind of guardrails might help?

said Charlotte Burroughs, chair of the EEOC session on Tuesday Entitled “Navigating Job Discrimination in Artificial Intelligence and Automated Systems: New Frontiers for Civil Rights,” it is part of a larger agency initiative looking at how technology can be used to recruit and employ people.

She said everyone needs to talk about the controversy surrounding these technologies.

“The stakes are simply too high to leave this to the experts,” Burroughs said.

Resume scanners, chatbots, and video interviews may create bias

last year , The EEOC has issued some directives On the use of sophisticated recruitment tools, noting its many shortcomings.

Resume scanners that prioritize keywords, “virtual assistants” or “chatbots” that sort candidates based on a set of predefined requirements, and software that evaluates a candidate’s facial expressions and speech patterns in video interviews The agency found that it could perpetuate prejudice or create discrimination.

Take, for example, a video interview that analyzes the applicant’s speech patterns in order to determine his problem-solving ability. A person with a speech impediment may receive a low score and be automatically eliminated.

Or a chatbot programmed to reject job applicants with holes in their resume. The bot might automatically reject an eligible candidate who had to stop working because of disability treatment or because they took time off to deliver a baby.

Testifying during the hearing, Heather Tinsley Fix, senior counsel for the American Association for Scientific Research, said that older workers may be disadvantaged by AI-based tools in multiple ways.

Companies that use algorithms to scrape data from social media and digital profiles may ignore professionals in the search for these “perfect candidates.” who have smaller digital footprints.

Also, she said, there’s machine learning, which can create a feedback loop that then harms future applicants.

“If an older candidate makes it through a resume screening process, but gets confused by a chatbot or reacts poorly, that data can tell the algorithm that candidates with similar profiles should be ranked lower,” she said. .

It can be difficult to know that you have been discriminated against

The problem is for the EEOC to root out discrimination — or prevent it from happening — when it’s buried deep within an algorithm. Those denied employment may not connect the dots with discrimination based on age, race, or disability status.

In a lawsuit brought by the Equal Employment Opportunity Commission (EEOC), a woman who applied for a job at a tutoring company realized the company had set an age limit after she reapplied for the same job, providing a different date of birth.

The EEOC studies the most appropriate ways to deal with the problem.

On Tuesday, members of the panel, a group that included computer scientists, civil rights advocates and employment attorneys, agreed that audits are necessary to ensure that software companies use avoids intentional or unintended biases. But who will conduct these audits – the government, the companies themselves or a third party – is a thorny question.

Burroughs noted that every option presents risks. A third party may be outsourced to treat its customers leniently, while government-led scrutiny could stifle innovation.

It was also discussed setting standards for vendors and requiring companies to disclose the recruitment tools they use. What these will look like in practice is not yet clear.

In earlier remarks, Burroughs has pointed out the great potential that algorithmic decision-making tools and artificial intelligence have to improve the lives of Americans, when used properly.

“We must work to ensure that these new technologies do not become a high-tech route to discrimination,” she said.