Connecting state and local government leaders
Utah simplifies the process of registering brands using image recognition technology, one of many projects that is part of the state’s AI plan.
The state of Utah has combined one of humans’ oldest practices—branding livestock to identify the owner—with one of the newest technologies: artificial intelligence.
Ranchers have registered 16,000 brands with Utah’s Department of Agriculture. Previously, when a new one came in, employees would use keywords to search through descriptions of existing brands to see if the proposal could infringe on a mark already used by another rancher.
“All of a sudden, there is now a simpler way to do it,” said Mike Hussey, Utah’s chief information officer. Using image recognition technology, the agriculture department worker can more easily search the state’s database to see if the new brand conflicts with existing ones.
On the flip side, if a cow or bull escapes and needs to be identified, a photo of the brand on the animal can be more easily plugged into the system to find the owner, Hussey said.
For agriculture department workers, it’s a time saver. They “can now do more rewarding work,” he said.
Hussey said the larger promise of the branding program—which was developed this year using money from the state’s Department of Technology Services’ innovation fund—is in deploying the image recognition tech to other agencies.
“We built the skill set in house. We know how to touch some of these cloud providers that have the image recognition in place,” he said. “And now we can just build it into other aspects of state government.”
Utah has long been a leader among states adopting AI technology, initially rolling out applications aimed at residents, like one to help find fishing spots and an Alexa service to help teenagers study for the driver’s license exam. Dave Fletcher, the state’s chief technology officer, said those initiatives showed leaders throughout government the potential of adopting AI. In 2019, the state created an Artificial Intelligence Center of Excellence, which has been the engine for more recent work.
“We focus on technology that we think can have the greatest impact on government. What technology can improve efficiency, how can we do things better, how can we improve the delivery of government services,” Fletcher said. “When we looked at AI, it has that potential. It is really a whole suite of technologies.”
As in other states or localities, AI has also prompted ethical discussions, prompted at least in part by a controversy involving Banjo, a Utah-based company that had a contract with the state to tap into surveillance systems and deploy technology to detect crimes or other incidents. That initiative was suspended by Utah Attorney General Sean Reyes’ office after news reports that the company’s owner had previously been involved with the Ku Klux Klan.
But even before that story came out, some lawmakers and civil libertarians in the state had questioned whether the project violated people’s privacy.
Fletcher said one of the center of excellence’s initiatives this year is to work on developing ethical standards related to uses of artificial intelligence, including facial recognition.
Right now, Hussey said the only facial recognition system used in the state is a fraud prevention effort to check whether people’s new driver license photos match their old ones. “Some of the things out in the public space, you are worried about crossing lines and privacy. This is one thing where I think it applies and it is a safe boundary for states to look at,” he said.
In the early stages of the coronavirus pandemic, AI became an important tool to get out information through chatbots, Hussey said. It was used most importantly in the state’s unemployment application process at a time when people were desperate for information about how to apply and state employees were dealing with a large increase in applications.
“Offloading some of that work to the chatbots to help folks get the information they needed, know what they needed to do next, really helped,” Hussey said. “I think it took a lot of the workload off of some of the folks here at the state.”
Fletcher said the state also used AI through Chronicle, a cybersecurity company that is part of Google Cloud, as cyberattacks escalated when many state employees shifted to working from home.
“We really have to use those kinds of tools to be able to respond effectively and understand where the threats are coming from,” he said.
Hussey noted that the number of attacks rose from 1.4 billion to 2.3 billion daily, as hackers looked for new ways to get into the state’s systems.
“We shifted some of our log analysis onto a very robust engine in Google and some machine learning elements there that allowed us to get out in front of the threats that we were seeing,” he said. ”Humans can’t look at that kind of data, so you have these machine learning algorithms that will look and help the state shore up its defenses.”
Laura Maggi is the managing editor of Route Fifty.