NC enlists chatbots for government service
- By Matt Leonard
- Oct 13, 2016
Many private-sector businesses have started building chatbots into their products to improve customer service. Facebook users interact with a chatbot to order flowers, and Slack has made it easy to use build and launch customized bots on its platform.
In government, where customer service sometimes must cover to the nation’s whole population, chatbots are being eyed as way to improve citizen-facing services and automate simple, repetitive employee tasks.
The North Carolina Innovation Center, which tests technology that will improve customer service and make government more customer friendly, wants to use chatbots to both provide basic IT help to employees and make it easier for citizens to reach the services they require.
When private tech companies started talking about chatbots earlier this year, North Carolina Chief Technology and Innovation Officer Eric Ellis became excited about the technology’s potential for state government.
A group at the Innovation Center has been looking at specific use cases for about three months, and Deante’ Tyler, the deputy director of the Innovation Center, said there are two main areas where chatbots could be used.
“First, we’re looking to, as a government, improve the lines of communication we have between the public sector and our citizens. Chatbots will be a new platform for engaging our citizens,” he explained. And second, chatbots can provide the self-service options that citizens and state employees alike desire.
The center’s first project, which Ellis said could go live within a few months, is a chatbot for employees who need IT help -- to recover lost or forgotten passwords or usernames, for example. The state gets thousands of simple tickets like this every day, and they would be easy to resolve with a chatbot that had the right information, Tyler said.
But the crew at the Innovation Center wants to be sure the technology is streamlined and thoroughly tested before launching any chatbot. “We’re just being cautious before rolling it out into a full project,” Ellis said.
This means finding the right platform and testing it. For internal use, the staff is working with Slack, which they think could be a good platform for the employee help desk, Ellis said. But for more citizen-facing applications, a mobile app or a web-accessible chatbot might be best.
The internal help desk chatbot will likely be rolled out first, Ellis said, because the Innovation Center wants to road-test the technology and ensure it can deliver a good user experience before putting it in front of citizens. This mean using natural language recognition technology and taking into account the different ways people could ask questions. While natural language processing might be more expensive, it is the “secret sauce” to making a chatbot work, Ellis said.
And any chatbot’s success will depend on the quality of the interactions, he said. Poor implementation could mean losing users, something they hope to avoid. As the Innovation Center experiments with the chatbot, the team will collect data that will help it fix problems at the source or just look at ways to improve the product, Tyler said.
Others have seen the potential for chatbots in government. At a June discussion hosted by the General Services Administration, federal agency and industry experts suggested bots could answer questions related to taxes or improve accessibility to services for people with disabilities or speakers of foreign languages.
Matt Leonard is a former reporter for GCN.