Government agencies nationwide are exploring ways to make digital services more inclusive, with efforts ranging from expanded focus on accessibility to investment in digital equity. Translation AI technologies are part of this work, with the education sector already seeing the impact.
The state of Minnesota has been involved in a multiyear language access effort to make DVS services more accessible to more Minnesotans. Translation AI has been implemented via a chatbot, or virtual assistant, launched earlier this year. The tool creates an avenue for constituents who do not speak English to interact independently with government, and its use may expand to include additional services in the future.
The tool’s interactive chat feature allows users to get information using multiple languages and to access services such as updating their insurance, checking the status of their title, or checking on their driver’s license directly within the chatbot, DVS Director Pong Xiong said. Users can get updates on the status of a service such as receiving their title throughout the process for more clarity.
The creation of this tool involved identifying services that could be effectively integrated in the first iteration, and the current priority for DVS is ensuring that the customer experience remains efficient as new services are integrated.
However, Xiong noted that future iterations may include additional services, such as enabling people to renew their registration or buying replacement stickers.
“We want to do more, and really, the key here is self-service,” he said, underlining the goal to make as many services as possible available any time or day.
The decision to use this tech emerged organically through an effort that really started in 2020 in a town hall listening session, during which DVS was able to gain insight from community members about service needs. It became clear to officials that a primary challenge for constituents interacting with DVS services was the language barrier.
More specifically, for DVS to meet the unique needs of citizens, it was important to find a tool that offered translation for Somali, Hmong and Spanish.
“Those three languages make up over 80 percent of non-English speaking languages in Minnesota, and so we know that these languages get a critical mass for us,” Xiong said.
Prior to the chatbot’s launch, Xiong said that customers who did not speak English could navigate DVS services using a phone line with translation services or using their own translators or family members for support. He underlined that one of the values of this tech is the human piece of the experience, because it empowers constituents to navigate government services independently.
The communication with community members has been a major piece of this tool’s development. Xiong highlighted the value of engaging early and often — from understanding the needs of Somali, Hmong and Spanish communities, to collaborating with working groups to test the tools.
He also noted that Hmong is a young written language, which amounts to a limited amount of content for the machine to take in. This can make translation of common terms more difficult, and especially regarding technical jargon for DVS-specific language.
“That’s something that we didn’t quite anticipate going into this, but we’re absolutely committed because that’s where our customers are at, and that’s what we need to do for Minnesotans.”
The state is working directly with the Hmong-speaking community and obtaining feedback and will continue to do so through the ongoing process of building out this database for continually improved translations.
The chatbot, powered by Google AI, builds on a previous DVS chatbot, the TAMI bot, which offered formulaic responses based on keywords. With this technology, there is a more conversational flow that helps users get access to services more organically.
DVS is gathering some metadata through the chatbot, in addition to collecting feedback from customers about their experience, to improve the tool. This will allow DVS to see if more information should be added on specific topics or if there are gaps or issues with the tool.
Because of the nature of the AI landscape, regulatory requirements may evolve, but Xiong said the focus remains on balancing the use of AI to improve the customer experience while staying within accordance with any new policies put in place moving forward.
