Less than one year after OpenAI started getting momentum as a pure-play machine translation option for enterprises, the world’s most valuable AI startup revealed that it is now providing machine translation at scale for a public institution.

Featured as one of currently 47 so-called “stories” on the OpenAI website, Minnesota’s Enterprise Translation Office (ETO) outlined how it is using OpenAI’s technology to address “challenges” in providing translation services, which were “highly decentralized […] [and involved] engaging external contractors, leading to inconsistent quality, high costs, and delays that could extend up to a month for a single request.”

The ETO piloted the solution for four months before deploying a full rollout in July. OpenAI stated that the institution’s new translation workflow is now “more efficient, [and] sets a new standard for language inclusion in the public sector.”

In addition, the solution “optimizes for cultural relevance to ensure government messaging is both accessible and inclusive. For example, with Hmong and Somali, the ETO developed spreadsheets with culturally relevant terms and translations and incorporated these into their GPTs to improve the model’s cultural awareness.”

2024 Slator Pro Guide: Translation AI

The 2024 Slator Pro Guide presents 20 new and impactful ways that LLMs can be used to enhance translation workflows.

In a plot twist for local interpreters — who went on strike in January 2024 over pay for court interpreting — OpenAI also confirmed that the ETO is now “supporting a limited pilot project with another state agency to test the use of ChatGPT’s voice capabilities for real-time interpretation.”

The tech company stated that interpreting services in Minnesota were “equally challenging, where limited access to approved interpreters made it difficult for people to engage with critical legal and community services effectively.”

“[ETO is] supporting a limited pilot project with another state agency to test the use of ChatGPT’s voice capabilities for real-time interpretation.”  — OpenAI

The pilot — whose “early results are promising” — aims to improve accessibility for the state’s non-English-speaking citizens in various live situations to “provide faster, more efficient communication for those with limited English proficiency.”

In Minnesota, over 20 percent of local residents primarily speak a language other than English, with Spanish, Somali, and Hmong being the top three major non-English languages spoken in the state.