Today was one of the first meetings of Pennsylvania's artificial intelligence advisory committee. Industry leaders, researchers, and policy experts came together to discuss the growing need for state rules in this influential industry.

"There's a lot of things that are going on already in the administration and with artificial intelligence,"Rep. Bob Merski (D-Erie) said. "And I think there are some things legislatively that are reactive to things that are happening in the community and in the world at large.”

Merski is one of two elected lawmakers on the committee. He wants to prioritize protecting both productivity and privacy is key in any laws passed.

“Taking the 30,000 foot view, which is what we're doing— I think that's the way to make sure that you don't have unintended consequences to your legislation,” Merski said.

Only a few states have passed artificial intelligence laws, but many federal agencies have created and are enforcing ai guidelines.

"The Food and Drug Administration has regulations related to artificial intelligence in medical devices,” Cary Coglianese said, the director of University of Pennsylvania’s Program on Regulation.

Coglianese says creating regulations for artificial intelligence is challenging because of how its not a singular entity, but a concept with unique applications in many industries.

“It is a technologies, plural, and lots of them,” said Coglianese. "And they're also changing dramatically.

Because artificial intelligence is evolving rapidly, he says regulations should not be fixed guardrails, but more like a leash.

“What we need is an ai oversight that is flexible, that's agile and also vigilant. We can't have society just think that, well, we've passed a law and that's enough. This is an ongoing process, and we need to have government oversight that's going to be operate on a continuous cycle."

He proposes what he calls "management based regulations”, which rely on private sector entities to create personalized plans of regulation while still having government oversight on the legitimacy and quality of the plans.

On a broader societal conversation, he says hopes and fears of artificial intelligence being better or worse for humanity must be kept in perspective.

“Compared to what? Compared to the status quo, humans are not perfect either,” Coglianese said. "If we can do better with a in certain domains, then that's where regulators need to make sure that they're not standing in the way. But they also need to make sure that we try to minimize things from getting worse."