Utah’s tax commission AI assistant

The Utah Tax Commission is responsible for collecting, processing, and enforcing all the state’s taxes. As is the case with other states, this office is besieged with queries from residents who do not want to make a mistake on their filings. To deal with this, the state currently employs 200 call-center agents, with knowledge based on experience and training.

Since there’s a fair amount of turnover in these ranks, training is an ongoing process, and even then, the newer agents tend to be less prepared to respond quickly and accurately.

So, the state ran a pilot utilizing four vendors who used a large learning model to produce an AI tool that could map answers to the 366 typical questions that people ask when they call in, based on an initial software script analysis. Those answers were then analyzed by knowledgeable tax and subject matter experts to judge the result compared to what a knowledgeable call center agent would say.

The result: 92 percent of the time, at least one of the models answered the question as accurately as a knowledgeable agent. Vendors thought they could get closer to 99 percent by fine tuning.

The state picked one vendor to roll out the use of the AI tool in pilot form on a small number of agents, and the model is being further refined. “And we expect to roll it out to all of the agents to give every agent a very powerful digital assistant that can help them answer questions more accurately and faster,” said Utah’s CIO Alan Fuller.

The creation of an AI chatbot that could be accessed directly by the public is still a work in progress. “We’re aggressively working to get a chatbot put together,” said Fuller, “but we’re going to keep it internal until we get to the point where the tax commission feels confident that it is giving accurate enough and consistent enough answers that they can make it available directly to the public. If you come to the tax commission for help with your taxes, you need to be able to trust that you’re getting good information back.”

Attribution: Case study provided to the AI Navigator courtesy of the IBM Centre for The Business of Government, Apolitical's content partner. “AI in State Government: Balancing Innovation, Efficiency, and Risk”, https://www.businessofgovernment.org/reports/ai-in-state-government, IBM Centre for The Business of Government.