For several years, the state of Utah was collecting statistics and feedback on public opinion, but the state didn’t really have a plan for what to do with the data.
Recently, it decided to use machine learning tools to analyze health, transportation, air quality and geo-based Twitter information to perform sentiment analysis before, during and after Utah’s winter inversions and air quality spikes. Utah CIO Michael Hussey explained how the state went about it at the 2018 National Association of State Chief Information Officers (NASCIO) Midyear Conference in Baltimore on Tuesday.
Winter inversions in Utah occur when the usual atmospheric conditions become inverted. A dense layer of cold air becomes trapped under a layer of warm air, essentially sealing pollutants closer to the ground. Extended inversions can lead to unhealthy levels of fine particulate pollution, which can result in poor air quality and in turn can affect the health of residents.
By performing an analysis of the public outlook on factors that may contribute to poor health during winter inversions, Utah could pinpoint trouble areas and seek policy solutions tailored to those areas, according to Hussey.
Hussey said Utah is exploring the use of machine learning in several cases. The state uses a digital assistant to provide practice exams for the state driver’s license test and is exploring doing the same for a notary exam. Meanwhile, Utah residents and visitors can visit ontime.utah.gov for real-time transit information.
Utah and other states might see additional applications for machine learning in fraud detection, interacting with the public using chatbots, image processing, cybersecurity automated with artificial intelligence (AI) and sensor data connected to the Internet of Things, Hussey said.
“Machine learning makes a difference when you coordinate high volumes of data on the fly,” he said.
Intelligent Automation Will Make Government Smarter
Speaking at the same NASCIO panel, Gartner Research Vice President Rick Howard observed that government will become more conversational and location-aware in the near future, and it will offer “Anything as a Service,” empowering more and more programs through the internet.
“Any one of those things by itself may not be remarkable, but combined these things are transformational,” Howard said. “It will be about taking your data assets and making machines smarter, your services smarter and government more intelligent.”
Government is at an early stage of AI adoption, Howard said, but 50 percent of top performers in global government are already implementing AI, according to Gartner survey results. For state and provincial governments, 21 percent are using AI, primarily in fraud detection.
Promising AI use cases include personal assistants, virtual call centers, human resources applications such as resume screening, anomaly or fraud detection, opinion-mining analyses and customer segmentation for marketing, Howard said.
Howard suggested that a better way to think of artificial intelligence in these scenarios was to regard it as “intelligent automation,” applying machine learning to automate repetitive tasks, particularly high-volume repetitive tasks.
When considering whether AI can help quickly, states should consider whether a business opportunity has a sufficient amount of quality data that can be understood and has generally agreed upon choices or outcomes. If so, governments should pursue AI applications for those opportunities. But if not, they should develop or acquire data science skills to leverage AI solutions or postpone consideration, Howard said.
“If you’re more than three years behind implementing commercial technologies for your citizens, you are falling behind,” Howard said.
Bookmark this page to follow StateTech’s coverage from the NASCIO 2018 Midyear Conference. Follow us on Twitter at @StateTech and keep up with the latest conference conversations using hashtag #NASCIO18.