That includes chatbots, which governments have used to help handle a massive influx of questions from the public. The idea is that chatbots, which typically use some form of AI algorithm, can handle common questions and leave less common or more complicated questions for human staff to answer.
And while the data on chatbot use shouldn’t be treated as a nationally representative sample, it does provide a picture of the general state of chatbots in government.
Most jurisdictions that use bots have a definite list of questions they are capable of answering — in other words, we aren’t talking about Asimov-style intelligences that can learn to solve new problems and answer new questions on their own. Rather, the IT department might spend time configuring the chatbot before it goes live in order to work on how the bot recognizes and responds to questions.
They are often structured for triage, the weeding out of people whose questions can be answered easily so that call-takers can focus on people whose questions will take more effort. That’s especially important during the pandemic because, as many vendors and governments have documented, people have been turning to the government a lot more than usual for things like health testing, unemployment benefits and other kinds of assistance. Records have been set all over the country for visits to .gov websites, as in-person office visits became restricted, placing new emphasis on all citizen engagement channels.
In King County, Wash., for example, a chatbot helped identify which people calling for a professional nurse had coronavirus-like symptoms and which didn’t. The county estimated that the chatbot saved 35 percent of the time nurses had been spending speaking to people without those symptoms.
And in states across the country, chatbots served a vital role in augmenting the capabilities of human staff members to deal with unrelenting waves of questions from applicants of unemployment programs. The Texas Workforce Commission, for example, relied upon chatbots to field frequently asked questions, again letting employees focus on more complex issues.
A lot of the jurisdictions surveyed used their chatbots for COVID-19-related purposes. Connecticut’s COVID chatbot, for example, built using technology from IBM Watson, logged nearly 40,000 interactions in a four-month period beginning last March. The state estimates that it did the work of four full-time employees during that time. But chatbots often proved useful well beyond COVID-19 needs as well.
Placer County, Calif., for example, has a bot called Ask Placer capable of answering more than 375 questions. IT agencies in San Joaquin County, Calif., and Fairfax County, Va., both worked with other departments to figure out what their needs were and what their most frequent questions were so that they could build those into their chatbots.
Minnesota has a similar approach, leaning on its IBM Watson chatbot to help address general inquiries. Iowa’s chatbot dates back to late 2018, and capabilities continue to be added as new needs arise. Seventeen agencies now use it, and so does the public. In May 2020, the state’s chatbot tools, combined with its live chat function, saved an estimated 1,700 hours of staff time that would have been spent addressing those same inquiries using traditional tools.
Cabarrus County, N.C.’s chatbot was integrated with Laserfiche technology in order to help people use digital services. The chatbot is capable of pulling in information from other systems in order to help the user. Missouri’s Department of Revenue worked with Accenture on a virtual agent named DORA, which answered 100,000 resident questions in its first three months since its debut in November 2019. Agency goals were to help field questions on taxes, driver’s licenses and motor vehicles, though the tool has also proven useful in addressing pandemic and unemployment-related inquiries. Since digital services have become a necessity for many government agencies that have found themselves transitioning to telework during the pandemic, tools to help citizens use digital services make a lot of sense.
Caroline, South Carolina’s chatbot launched in early 2020, made it that much easier for the state to stand-up a COVID-specific chatbot later in the year.
A key feature of chatbots is that they’re designed to answer a growing number of questions over time. Kansas City’s Facebook chatbot was born several years ago out of a desire to make it easier for citizens to navigate and use its open data portal. The effort at the time got a lot of grassroots help from the local Code for America Brigade, and city officials considered the chatbot an experiment of sorts, to better understand how the technology could be used more broadly.
Many governments use data analysis tools to follow the kinds of questions citizens ask — as well as the ways they ask them — so that they can add answers to those questions over time, and so bots can learn how to respond to variations. San Jose, Calif., for example, has a chatbot on its 311 page, SJ311, which gathers feedback to fuel continuous improvement.
South Carolina’s first foray into chatbots was Caroline, developed with its partner NIC and launched in early 2020 to provide a simple path to information for site visitors. The deployment undoubtedly smoothed the way in mid-2020 for Axel, a chatbot feature incorporated into the state’s COVID-19 recovery portal, accelerateSC.
Chatbots can also take inputs in many different forms, which gives them the unique ability to serve citizens across multiple channels. Several jurisdictions got into chatbots by first making them available via text, a more ubiquitous option a few years ago when the technology was first taking off. This was the case in North Charleston, S.C., as well as Williamsburg, Va., which this year has added Web functionality, greatly increasing citizen use of the chatbot. Washington County, Ark., too, is working on a bot that works with texting, while Placer and King counties integrated with Google Assistant and Amazon Alexa so residents could access the bots with their voices — a practice also gaining ground in government.
San Joaquin County also built its bot to work in three languages, with more planned for the future. This tactic, too, is spreading across government as jurisdictions seek to develop tools that are as useful as possible to the specific needs of their community. Bellevue, Wash., for instance, has a chatbot offering COVID-related information in seven languages.
Survey results and the trends of government technology during the pandemic point to a time of growth for government chatbots. Especially if they can help make digital services, emergency operations and telework more workable for local governments, their usefulness might make them hard for many jurisdictions to ignore.
Editor Noelle Knell contributed to this story.
*The Center for Digital Government is part of e.Republic, Government Technology's parent company.