AUSTIN, Texas — When Rep. Giovanni Capriglione (R-Southlake) asked state agencies to document how they’re using AI, he was a bit surprised by what came back.
“There were almost 700 different use cases already in state government, that are using artificial intelligence," Capriglione said.
For the state representative from Southlake, it was further proof of why his new committee is needed. “It’s not just about how private companies use AI, but also how the government is using it. Which could be just as worrisome, right?” he says.
Capriglione co-chairs the state’s Artificial Intelligence Advisory Council. And for the past 8 months, they’ve been doing a deep dive into AI in Texas. They asked experts to use AI tools to create deep fakes, to get a sense of the worst-case scenario.
“I asked them to come to the committee and do one against me. It just pointed to my LinkedIn, that’s all it had to do," Capriglione said. "It immediately finds out I’m in Tarrant County, and the Fort Worth Star-Telegram is the paper of record. It creates an article. Same font, same everything. It knows I live in Southlake so it puts ‘Southlake City Hall’ and creates an active shooter alert.”
The social post was fake but certainly didn’t look or sound like it.
The post read “Authorities are investigating multiple yet unconfirmed claims of an armed individual near voting sites in Southlake. Due to the security concerns, most polling locations will close early today.”
The co-founder of Civic AI Security Program, Lucas Hansen, asked the committee, “Imagine that goes out in the middle of election day?”
"While multiple agencies are using artificial intelligence, most of it is focused on cyber security right now," Capriglione said. "We’re being attacked all the time, here in Texas. And so we’re using AI to automatically protect us from these attacks.”
Some of the other uses are more mundane.
“I’ll give you an example,” Capriglione said. “TXDot does a lot of invoicing. Tens of thousands of invoices to pay contractors. It was very manual and used to take about three weeks to look at all the contracts and do the invoicing. Well, now it takes 27 seconds. So it’s a massive savings in time.”
TxDOT also launched a pilot program near Austin, integrating AI with web cameras to look at traffic patterns.
“By having artificial intelligence look at those video screens, they’re able to see accidents in real-time," Capriglione said. "It might’ve taken a while for a tow truck to show up. And what they’ve shared with us is, they’ve already seen a 5 to 10-minute reduction in traffic. And I expect that’s something that’s going to get rolled out [statewide.]”
Up until now, perhaps no state has more than California when it comes to implementing and legislating artificial intelligence. The city of Stockton has installed AI cameras on its code compliance trucks, to detect violators. And several bills are on Governor Gavin Newsom’s desk, waiting to be signed into law.
One requires you to get consent before using AI to replicate a dead performer’s image. Another mandates community college classes be taught by human professors, not AI. And the most controversial would hold developers liable for how certain high-powered AI systems are used or modified.
“I think that some of the laws that other states have worked on are probably too heavy-handed in some ways,” Capriglione said. “When it comes to regulating AI, I think Texas is going to do it in a very smart, rational way.”
He says Texas is trying to strike a balance between protecting residents, without strangling the innovation AI can bring.
“There are definite corporate interests," Capriglione said. "We want to make sure that rather than strictly follow corporate interest, we’re following human values.”
But the promise of technology this powerful carries an implied threat, Capriglione said.
“We want this to be useful to us," he added. We want it to work for us. We don’t want to work for it.”
There’s only so much Texas can do to manage AI’s evolution, and nothing it can do to stop it. When Civic AI Security Program’s Hansen spoke to the committee earlier this year, he posed this question: “What do artists, programmers, lawyers, journalists, customer service reps, teachers, translators and therapists all have in common? Well, it seems like AI can do a significant portion of their job. Maybe not the entire thing, but a lot of it.”
Capriglione doesn’t see Texas going the route of California, in terms of legislating that certain positions must be protected.
“Is it going to disrupt jobs? Yes. But what does that mean? Technology has always had disruptive effects on the economy," Capriglione said. "So yes, I think, and I believe most people would agree, this is going to get rid of some jobs. And what Texas must do, and the new companies coming in, is to ensure that our workforce can maintain the skill to find the new jobs.”
Capriglione sees the Advisory Council’s role as giving Texas a road map to navigate this new world. They must compile a report by December 1st, telling state lawmakers how agencies are using artificial intelligence right now, whether Texas needs an AI Code of Ethics and recommending certain policies to the 89th Legislative Session, which begins in January.
“And as this technology is no longer just something people use to create documents but involved in every single facet of our constituents’ lives, it has to be part of the whole policy conversations we continuously have,” Capriglione said.
Remember that list of professions that Lucas Hansen rattled off, where he singled out certain jobs that even the current version of generative AI can somewhat replicate? Capriglione says he missed one: state lawmaker.
“You’re going to see laws passed with bills completely written by AI,” Capriglione said.