Lawmakers say artificial intelligence is coming to state government

BATON ROUGE, La. (WAFB) - Lawmakers at the state capitol are trying to understand how to use artificial intelligence and what guardrails are needed to ensure personal data is secure. The Joint Legislative Committee on Technology and Cybersecurity met today to hear from leaders in the A.I. space. Derek Williams, Chief Information Officer at the Louisiana Office of Technology Services said we need to tread lightly.

"There's trillions, trillions with a t, of dollars going into this right now," Williams said. "So yeah for the state agencies, the biggest concern for me is people using these free tools without guidelines around them. That's where we're going to fall down."

A main concern from the committee was state agencies using automated decision-making tools from A.I. that could determine a person's eligibility for government services.

"As long as we're not using A.I. to make big critical decisions to determine someone's benefits for Medicare and Medicaid, or snap benefits, or something like that to actually make those decisions. Using it in an assistive way to generate or view that then has a human element that is doing the oversight."

In addition to a call for oversight, the committee also tried to clear up misconceptions about using A.I. models, especially when it involves personal information. Graham Ryan, a lawyer who works in the A.I. space said this is a big risk when people are ill-informed."If you're using a free A.I. tool and your prompting input, likely, whatever you put into that generative AI system is no longer private," Ryan said. "That could be used by the company to maybe retrain its model and use in something else in the future."

Williams said big tech companies have already thought about solutions to this problem.

"What people have started doing like Google and Microsoft and others is you can get copies of the trained model and you can basically enter into a contract to pay them to make sure that the data you give them does not get used to train again later," Williams said.

Right now, there are over 120 bills in the United States Congress to address issues with A.I. From election interference to pornography.

However, none of those bills have passed, leaving the states to come up with their own policy. Leaders in the space, like the company TechNet, are asking the legislature to be specific in their regulation.