Artificial intelligence is increasingly woven into our lives, but civil libertarians say that if it is improperly used, it can perpetuate historic discrimination and make mistakes that can dramatically alter someone’s life. (David Espejo/Stock photo via Getty Images)
The proliferation of decision-making artificial intelligence systems into consequential facets of life is posing threats to Americans and increasing racial biases, tech and legal experts with the American Civil Liberties Union said during the organization’s AI Summit Thursday.
“We’re now at a place where automated systems and AI are being used increasingly throughout our lives, determining who gets a job, who gets a loan, who gets policed and surveilled, and more,” said Marissa Gerchick, a data scientist and algorithmic justice specialist with the ACLU.
Decision-making algorithms analyze and sort through data, find patterns and make predictions or recommendations. They were built to cut down the amount of time this work is done by humans, and are being widely used across industries — 98% of Fortune 500 companies now use AI tools in their hiring process.
But imperfect training practices and deployment of these AI tools can create discriminatory outcomes, which hurt people’s economic opportunities or freedoms, ACLU panelists said.
“These systems are really built on prior human decisions that show real evidence of discrimination,” ReNika Moore, director of the ACLU’s racial justice program, said.
Moore said the ACLU often sees AI in the early stages of hiring, like ad targeting and candidate screenings. Moore said companies often use AI to rank candidates or use it for their initial contact. The ACLU is working on a few cases relating to discrimination through AI-based hiring practices, she said.
In one, an autistic man was not moved forward in the hiring process after an AI system asked him questions similar to a clinical assessment for autism or other neurodivergent conditions. In another case, a Native American woman who is deaf was not offered accommodation for a video-based interview run by AI. Both complaints allege that the hiring systems violate Title VII and the Americans with Disabilities Act, Moore said.
Although AI algorithms have become more widespread in recent years, they’ve been used for decades, and they’ve had a history of discriminatory results, Moore said. An AI program meant to dole out money to rebuild homes after Hurricane Katrina awarded identical homes in predominantly white zip codes more money to rebuild than those in predominantly Black neighborhoods, she said.
“That’s an example where even if the data is correct in terms of the market values of the homes, the way the system was set up to deliver relief and determine relief packages would still result in any unequal outcomes,” Moore said.
Policing is another area where incorrect use or outcomes by algorithms poses life-changing and even life-threatening results, said Nathan Freed Wessler, a deputy director of the ACLU’s Speech, Privacy and Technology project.
Facial recognition technology (FRT), which assesses a photo, often called a probe image, against a database of public photos to look for a biometric match, is increasingly being used by police to solve crimes, Wessler said. But without proper investigative tactics, the use of FRT has led to wrongful arrests.
“They’re probabilistic systems, they’re not designed to make an identification, they’re designed to find similar looking faces,” Wessler said. “And so it’s no surprise that we’ve seen a number of cases of wrongful arrests around the country, almost all of which have involved wrongful arrest of Black people after police have relied on what turned out to be incorrect results of this technology.”
Wessler worked on the case of Robert Williams, a Farmington Hills, Michigan resident who was wrongfully arrested on robbery charges after FRT matched security footage of the incident with an expired driver’s license photo in the state police database. Detectives didn’t bother to verify his whereabouts before making the arrest.
The ACLU won its lawsuit against the city of Detroit, which required it to change how the police department uses the technology. Police can no longer go straight from facial recognition technology results to a witness identification, and they cannot apply for an arrest warrant based solely on the results of a facial recognition technology database.
FRT is facing state-by-state legislation, as are many other uses of decision-making AI. The ACLU’s ultimate goal is legislation at the federal level that would provide baseline civil liberties rights for AI use for all people in the United States, and allowing the states to build on it, said Cody Venzke, senior policy counsel in the ACLU’s National Political Advocacy Department.
The ACLU pursues policy options in other ways, too, Venzke said, like working with federal organizations which have significant authority over new technologies. During the Biden administration, they worked with the Office of Management and Budget on guidelines to responsibly use AI in government work.
Policies that promote transparency in AI could help cut down on discriminatory instances, the panelists said. When people know that they are interacting with AI, it allows them to better assert their rights, or ask for what’s legally available to them, Moore said.
It can be hard to know if or how discrimination has taken place without the transparency on what AI was used for, or how the algorithm works, Venzke said.
In 2023, Congress introduced the AI Disclosure Act, which would require AI users to include a disclaimer that content was generated by AI. Though the bill didn’t progress through Congress, many states have passed similar laws, especially relating to disclosure of AI in hiring.
“We can work towards shaping a better future in terms of what technology looks like and the role that it plays in our lives,” Gerchick said. “To sort of recognize that in a digital age, digital rights have to be part of our civil rights.”
This post was originally authored and published by Paige Gross from Missouri Independent via RSS Feed. Join today to get your news feed on Nationwide Report®.
