Artificial Intelligence Chatbot Debuting on

Jul 20, 2018

By Josh Moody

Employees in government offices answer hundreds of questions from citizens each day. Often, the same questions are asked over and over again and the same answer is given each time. Having someone sitting in an office answering the same questions every day isn’t very efficient.

Wouldn’t it be better to offload this type of work to a computer?

You can ask your smart devices to perform simple tasks for you (Hey Google, play my 80’s mix on Spotify). Why shouldn’t you be able to interact with Government in the same way? Enter the transforming technologies of Artificial Intelligence (AI) and Natural Language Understanding (NLU).

We all know what AI is. You’ve seen the “Terminator” movies, right? It’s kind of like that, but with less Arnold. AI is a huge field with many different interpretations. For simplicity, let’s define AI in this context as “a computer making decisions based on available information.”

The term NLU is not as familiar. It’s a specific area of computer science and artificial intelligence concerned with how humans and computers interact. Computers have traditionally been very rigid in their ability to process instructions. You had to ask the question in just the right way to get the answer you were looking for. But humans are less formal than that. We may have a dozen different ways of asking the same question. With our human intelligence, other humans can usually make sense of all forms of the question and understand the intent behind the question. Until recently, this was beyond the ability of most computer systems. With recent advances in AI and NLU, computers are increasingly better equipped to understand all the various ways humans give instructions and make decisions about how to carry out the command.

The INA team has been hard at work building the next generation of AI powered chatbot software to help citizens find answers to common questions. Using a list of frequently asked questions (FAQs), we direct the bot to the FAQs and have it figure out the most likely match for the question the citizen is asking. Does the bot get it right every time? The short answer is no, but there is a solution to get better responses from the bot called Machine Learning. We monitor the responses given out by the bot and when it gives an incorrect response, we train it by directing it to the appropriate response. The next time a similar question is asked, the bot will be a little better at answering. Over time, the bot will learn all the ways a question might be asked and the expected response. If the chatbot is asked something for which it doesn’t know the answer, it will hand off the question to a live operator for further assistance.

INA recently began piloting this program on Once we feel confident about the quality of answers the bot is providing, we plan to make this service available to our local and state government partners. Stay tuned for more developments in this exciting new area!


Written By Josh Moody

Josh joined INA in 2005 as a Senior Developer and now serves as the Director of Technology. He enjoys taking aerial photography with his drone and loves both Star Trek and Star Wars equally.


Information Network of Arkansas

Explore. Advocate. Simplify