States bullish on chatbots, report finds
Connecting state and local government leaders
Digital assistants helped states deliver critical information to constituents during the pandemic.
State governments’ use of artificial intelligence tools such as chatbots increased last year, according to a new report that dubs 2020 the “Year of the Bot.”
Overall, state officials are bullish on the technology, largely thanks to its role in addressing COVID-related challenges, states “AI Meets the Moment,” a report that the National Association of State Chief Information Officers (NASCIO), the Center for Digital Government (CDG) and IBM released Oct. 26. For instance, 60% of the 48 agency leaders and CIOs from 38 states who responded to the survey said they started using digital assistants, such as chatbots, to deliver critical information to constituents.
Sixty-five percent of respondents said that the AI tools they put in place to address the pandemic have delivered results as promised, with 4% saying they exceeded expectations and 18% saying they underperformed or were significantly challenged.
The survey compared AI adoption to the last time NASCIO canvassed states on their use of the technology in 2019 and found that while only 1% said AI was widely used across the state two years ago, 7% said it is now. In 2019, 13% said AI was in use but not a core line of business, compared to 60% now.
“The survey finds AI is driving significant results in areas such as improved service delivery and enhanced interactions with residents and constituents,” the report states. “However, despite improvements in decision-making and employee productivity, only 13 percent of respondents said AI generated cost savings. This underscores the challenges in establishing return on investment (ROI) in AI projects.”
The report points to Utah as “a leader among states taking beginning steps toward AI adoption.” It has put in place initiatives for connected vehicles, image recognition for cattle branding and sensors that analyze air quality.
“I think we’re at the point of inflection here on some exciting things,” Utah’s Director of IT Services Michael Hussey said in the report. “I think you’re going to see an explosion in the not-too-distant future.”
Arizona’s Department of Economic Security rolled out an inward-facing chatbot to help employees find what they need faster as unemployment insurance claims poured in. “For them to be able to type questions into that chatbot and get referred to the appropriate sections and references was very helpful,” J.R. Sloan, state CIO, said in the report.
In Texas, an AI Center of Excellence is running more than 24 programs to test AI and plans to develop sandbox environments to help with training and hands-on experience.
Asked what they need to support AI in the long term, 75% of respondents said a clear framework for use and governance, followed by a defined AI vision and strategy, a clearer understanding of vendor capabilities and a centralized approach to its adoption.
The biggest bottleneck to AI adoption is a lack of skilled staff training in AI, according to 79% of respondents. Eighty-five percent said understanding agency or department use cases and workflows is most important for getting the most value out of AI. Machine learning (ML) and data analysis skills followed at 58%, and data engineering rounded out the top three at 44%.
Sixty-five percent of respondents named legacy infrastructure the biggest bottleneck, and nearly half cited difficulties in identifying use cases. Privacy and ethical concerns ranked low.
Still, “optimism on AI’s potential to transform organizations long term runs high,” the report states. Confidence is strong in the next 12 to 18 months, with 63% of respondents saying they plan to use robotic process automation and nearly as many saying they plan to use ML. More than half plan to deploy digital assistants and natural language processing. Overall, 23% of respondents said AI has the potential to transform their organization now and 56% said it would within one to three years.
Call centers and data analytics tied for the top business processes states are considering for AI applications, followed by cybersecurity and health and human services.
States can take advantage of this early stage of adoption and the momentum to “lay a firm foundation grounded in sound strategy and savvy tactics” for AI, the report states. “Agencies must look at high-volume processes to see where there may be an opportunity for improvement, and then take a tactical approach to redefine the business process.”