When writing the nation's most sweeping AI law, Colorado focused on fairness and preventing bias: NPR

This spring, Colorado passed the the country's first comprehensive law about how companies and governments use artificial intelligence to make important decisions about people's lives.

“Whether (people) get insurance, or what the rate for their insurance is, or legal decisions or employment decisions, whether you get fired or hired, could depend on an AI algorithm,” warns Democratic State Rep. Brianna Titone , one of the most important legislatures sponsors of the bill.

The law does not target counterfeits or fraud, which some states, including Colorado, have cracked down on in other laws, but applies to the way AI is used in assessing people on things like college applications, hiring, loans, access to health care or insurance.

It will come into effect in 2026 and require companies and some government agencies to inform people when an AI system is being used. If someone feels the technology has treated them unfairly, the law allows them to correct the data they are using or file a complaint. It sets up a process to investigate bad actors.

“If you were fired by an AI process and you say, 'Well, this is impossible, there's no way I should be fired because of this,'” Titone said, “then you can get through the attorney general's office find a solution to say, 'We need someone to step in and check that this process has not actually caused discrimination and prejudice against that person.'”

She said that in some cases, AI has been found to give people an advantage based on their name or hobbies, such as, “if your name is Jared and you played lacrosse.”

Democratic State Rep. Manny Rutinel, another sponsor, said some provisions require companies to identify how algorithms can lead to discrimination and disclose how the data is used to train the systems.

“We still have a lot to do,” Rutinel said. “But I think this is a great first step, a really important and robust first step in making sure that technology works for everyone, not just a privileged few.”

Colorado's move is being watched by other states

The Colorado law grew out of a similar proposal introduced in Connecticut earlier this year that failed to pass. Other places have adopted more restrictive policies. New York City requires employers using AI technologies to conduct independent “bias audits” on some software tools and share them publicly.

“So states are clearly looking to each other to see how they can put their own stamp on regulations,” said Helena Almeida, vice president and managing counsel of ADP, which develops AI payroll services for a number of major companies.

“It will certainly have an impact on all employers and operators of AI systems,” says Almeida of the Colorado bill.

Matt Scherer, an attorney at the Center for Democracy and Technology, said companies have been using various automated systems, not even called AI, to make employment decisions for at least eight years.

“We really have so little insight into how companies use AI to decide who gets jobs, who gets promoted, who gets access to an apartment or a mortgage or a house or healthcare. And that is a situation that is simply not sustainable because, again, these decisions have crucial aspects that have a major impact on people's lives,” he said.

But he is concerned that Colorado law does not give individuals a specific right to sue for AI-related damages.

“There is definitely a lot of concern among unions and community organizations that this bill doesn't have enough power to really force companies to change their practices.”

There are already plans to change the law – it's just a start

When Democratic Gov. Jared Polis signed SB24-205 in May, he told lawmakers he was doing so with reservations. He wrote: “I am concerned about the impact this law could have on an industry that drives critical technological advances in our state for consumers and consumers. both companies.”

He said this is best decided by the federal government so that there is a national approach and a level playing field.

However, Polis said he hopes Colorado's law will advance discussion about AI, especially at the national level, and he asked lawmakers to refine it before it goes into effect. A state task force will meet in September and make recommendations in February. Polis has outlined areas of concern and asked them to focus regulations on software developers, rather than small businesses using AI systems.

Polis said the law could be used to target those who use AI, even if it is not intentionally discriminatory.

“I want to be clear in my goal to ensure that Colorado remains home to innovative technologies and that our consumers have full access to important AI-based products,” he wrote.

The industry is looking at this law and other laws that may be introduced

Michael Brent, of Boston Consulting Group, works with companies to develop and deploy AI systems to identify and limit the ways AI can harm communities.

“Companies have a desire to build faster, cheaper, more accurate, more reliable and less environmentally damaging systems,” he said. He said Colorado's law could promote transparency for people affected by AI.

“They can go into that space where they have that moment of critical reflection, and they can simply say to themselves, 'You know what? I actually don't want a machine learning system to process my data in this conversation. I would prefer to opt out by closing that window or calling a human if I can. ''

Despite all the focus on creating comprehensive regulations, Democratic Rep. Titone that Colorado is just beginning to figure this out with the tech industry.

“We need to be able to communicate and understand what these issues are and how they can be misused and abused.”

Bente Birkeland covers state government for CPR News.