Industry Experts Speak in Austin on AI’s Role in Government

AUSTIN — The Texas Legislature’s Innovation and Technology Caucus (IT Caucus) gathered industry experts on Tuesday at the REJ Conference Center to discuss artificial intelligence and its implementation in local and state government.

In a panel led by IT Caucus Chair Rep. Giovanni Capriglione (R-98), Texas CIO and Executive Director of the Department of Information Resources (DIR) Amanda Crawford, Texas 2036 Policy Advisor Rahul Sreenivasan and Google Practice Lead for AI and Machine Learning Amina Al Sherif navigated AI policy, development and implementation in state agencies as well as privacy and data concerns that come with its uses.

For the DIR’s part, Crawford mentioned both its award-winning AI Center of Excellence, launched in 2020, and announced the deployment of its Texas Technology Education and Innovation Center, a physical space for agencies and partners to safely navigate the integration of AI.

A consensus among the panelists was that the area in which AI can be the most effective is in back-office tasks, where there is less opportunity for public-facing system failures and more opportunity for human-AI task augmentation.

“I think there’s a lot of opportunity for taking AI and using it to automate a lot of back-office tasks that are extremely man-hour heavy, extremely labor intensive, and also frankly, not altogether super satisfying work when it comes to some of the numb work that needs to be done,” said Al Sherif. “Introducing AI into your organization is a transformative experience. This is not just buying a new CRM system. This is completely transforming the way you do business, from a people and culture aspect all the way to the type of tech that you touch. And I think that needs to be taken seriously insofar as not setting expectations that are too high in terms of us rushing to the finish line, but also understanding that with transformation comes a little bit of aches and pains.”

In concurrence with Al Sherif, Crawford said mundane tasks are where DIR is seeing AI used the most in state agencies, freeing employees to focus on more fulfilling and customer-facing tasks.

“If there ever was a place where we need more good humans who know how to interact with people, [it’s] in government,” said Crawford. “I look at our agency, at DIR, one of the things we do is review billing records from providers that come to us, and one of the services that we provide ... there are humans who go line by line by line by line over all of these bills to make sure that they’re accurate and that the state’s then billed appropriately. That’s something that we have automated a lot of if not all of it, but we’re looking to do more of that so the humans that we have working on those bills can go and do more customer service interaction and more customer fulfillment, where they’re actually working with agencies on problem solving and solutions instead of just looking at lines on lines on lines.”

Regarding AI’s potential to render certain roles obsolete, all three panelists stressed the importance of training employees in preparation for an increased reliance on AI.

Crawford noted that it is the people who know how to use AI that will ultimately replace those who do not rather than AI assuming the roles itself. Sreenivasan agreed, saying that while some jobs will become redundant, most jobs affected by AI will simply become different.

“I’m a lawyer, but I’m not necessarily worried about different jobs because there will still be a role for quality control,” said Sreenivasan. “Depending on what kind of prompt you put into ChatGPT, you’ll still need someone that understands statutes to make sure it’s giving the right outputs. Calculators and spreadsheets didn’t eliminate the need for accountants, so I think you’re talking about jobs being enhanced and changed, and it’s a matter of training up competencies and finding people that can work and deal with and use these tools in whatever new version of what their jobs will look like in the future.”

Still, Al Sherif said, AI will take some getting used to for many employees.

“You are telling someone who has maybe spent their entire professional career perfecting their craft and experience that now they need to pivot because the technology is changing things so fast,” said Al Sherif. “That’s extremely hard for anyone at any point in their career to do, whether you’re catching someone in their undergraduate studies or someone who’s very mature in their process of mastering their craft. So when it comes to aches and pains, I think it’s less of adopting and acquiring the technology and just kind of organizational operational motions you’d have to go through, and more of your employees are going to have to internalize and frankly work with you to figure out how automation can best make them faster and more efficient at their jobs, but also happier where they’re at.”

However, the DIR head argued that before AI can be fully integrated into all levels of state and local government, policy surrounding data privacy and protection must improve.

“Technology is not really the issue,” said Crawford. “It’s the policy, it’s the ethics, it’s the privacy, it’s the bias, it’s the protection of constitutional rights, it’s all of those things that are the hard part. With all candor, I don’t think that government gets it right and has a fully developed discipline around it.”

Sreenivasan agreed, attributing the issue to inconsistencies between existing systems and applications in individual agencies.

“I don’t think a lot of legislators and staff and policymakers understand just how fragmented existing data and IT systems are in our state government,” Sreenivasan added. “You could be at a single agency, and you might be shown some tool [where] there’s some proprietary applications being used to access the Unix database from like 1985 and they’re using that, but they can’t update their Windows 10 from a certain security update because that proprietary application won’t work on it. I mean, if that’s what you’re talking about, that’s not an agency, that’s not an infrastructure that’s ready to do anything that advanced.”