By Yonaira Rivera and Vivek Singh
In an increasingly digital world, chatbots have become essential tools for delivering health and government services. These trends are only expected to increase in the coming years. However, ensuring these services are equitable across different languages is a critical challenge. This blog post introduces the IMPACT-NJ project, which aims to address this issue by developing policy guidelines to enhance language equity in chatbot responses, particularly between English and Spanish, the two most spoken languages in New Jersey.
Language equity in chatbot responses is not just a matter of convenience; it is a fundamental aspect of ensuring fair access to essential services. In New Jersey, where a significant portion (~16%) of the population speaks Spanish [1], it is crucial that chatbots provide accurate, understandable, and culturally sensitive responses in both English and Spanish. This is particularly important in health and government services, where miscommunication can have serious consequences.
The IMPACT-NJ Project
IMPACT-NJ (Improved Policy-based Advancements for Chatbot Technology in New Jersey) seeks to elevate the standard of chatbot communication within New Jersey’s public sector. By learning from states currently leading in technological integration, such as Mississippi with its versatile chatbot MISSI and California’s PAIGE for IT procurement support, IMPACT-NJ aims to identify best practices that can be adapted to the New Jersey context.
The project will conduct audits of existing chatbot applications to assess their efficiency, inclusivity, and linguistic parity. This includes evaluating the disparities between English and Spanish responses and identifying areas for improvement. The ultimate goal is to create policy guidelines that ensure NJ-specific chatbots provide equitable responses in both languages, addressing the unique needs of New Jersey’s diverse population.
Insights from Previous Research
The IMPACT-NJ project builds on previous research conducted by the team, including studies on ChatGPT’s responses to health-related queries in multiple languages. For instance, a study comparing ChatGPT’s responses on vaccine hesitancy in English and Spanish found significant variations in the quality and hesitancy of responses across languages [2]. Such findings underscore the importance of cross-language evaluation to ensure equitable access to health information.
Policy Recommendations for New Jersey
Based on the insights gained from these audits and previous research, the IMPACT-NJ project will develop comprehensive policy recommendations. These guidelines will focus on ensuring that chatbots used in New Jersey’s public sector are capable of providing accurate, understandable, and culturally sensitive responses in both English and Spanish. This includes addressing issues such as readability, natural phrasing, and the avoidance of direct translations that may lead to misunderstandings.
Ensuring language equity in chatbot responses is essential for fostering informed health decisions and enhancing access to government services for all linguistic groups. By developing policy guidelines that promote linguistic parity, we aim to bridge the communication divide and ensure that all residents, regardless of their primary language, have fair access to essential services.
Yonaira Rivera is an assistant professor of communication at the Rutgers School of Communication and Information and Vivek Singh is an associate professor of library and information science at the Rutgers School of Communication and Information.
References:
- https://statisticalatlas.com/state/New-Jersey/Languages
- Joshi, S., Ha, E., Rivera, Y., & Singh, V. K. (2024). ChatGPT and Vaccine Hesitancy: A Comparison of English, Spanish, and French Responses Using a Validated Scale. AMIA Summits on Translational Science Proceedings, 2024, 266.