Considering a Generative AI Tool for Translation: Using a Chatbot for Multilingual School Communication
As a preschool teacher for 8 years, I see the changes from when I first started teaching. While I’ve always had two or three children whose primary language was Spanish, I’ve always been able to communicate with their families at least orally through my bilingual teacher’s aide and my basic Spanish from college. Now nearly half of my children are from families whose primary language is neither Spanish nor English. Gujarati, French Creole, Arabic, Bengali, Portuguese, and Mandarin are the principal languages spoken. Knowing the importance of teacher-family partnerships and communication, I’m at a loss as to what I can do.
As noted by this preschool teacher in New Jersey, the number of US families who speak languages other than English has dramatically increased in recent years. According to US Census Bureau data for 2018–2022, 21.7% of US residents speak a language other than English at home (United States Census Bureau 2022). While this has many societal implications, it is crucial for schools to consider as they aim to create positive, reciprocal relationships with all families. Studies have shown that families whose home language is not English experience obstacles to active participation (Subramaniam 2011).
The Role of AI in Multilingual School Communication
A critical way to strengthen family-school partnerships is through communication. To be effective, communication should come in many forms and happen regularly. The importance of providing flyers, letters, announcements, and school or program website information in multiple languages is well documented (Li et al. 2023; Piller, Bruzon, & Torsh et al. 2023). However, many administrators and educators are not adequately prepared to communicate with families who speak a language other than their own, due to a lack of time, resources, and guidance (e.g., García & Wei, 2014). Therefore, they may be in search of new or additional avenues to help facilitate family-school communication.
Exploring Generative AI for Translation
We (the authors) have extensive experience working with ethnically and linguistically diverse school districts and have observed that the increasing linguistic diversity in schools highlights the need for more frequent, consistent, and individualized communication in multiple languages to foster a more inclusive learning community. Various artificial intelligence (AI)-powered tools are currently available, including OpenAI ChatGPT, Microsoft Copilot, and Google Gemini. In this article, we focus on ChatGPT and its capabilities for translation into multiple languages. We outline steps for getting started with ChatGPT and crafting prompts for translating information for families. Finally, we describe some limitations that educators should be aware of when using AI-powered tools for translation.
Understanding Generative AI and ChatGPT
Artificial intelligence is “a field of computer science that refers to the theory and development of computer systems capable of performing tasks that mimic human cognitive processes” (Delello et al. 2024, 25). Generative AI refers to a type of AI that can create new content, such as text, images, and music, that is similar to the abundant amount of data it has been trained with (Delello et al. 2024, 4). Generative AI-powered tools such as ChatGPT are not designed for one specific purpose. Instead, they offer a wide range of possibilities for users to explore and utilize in different ways (Li, Kou, & Bonk 2023).
Educators may already be familiar with using tools that tap into AI (such as Google Translate) to create materials in multiple languages. Newer generative AI tools, like ChatGPT, have some advantages over apps like Google Translate. For example, they have the ability to decipher more about context and can generate multiple automated revisions. ChatGPT is a large language model-based chatbot, which is a computer program that imitates human conversations using text, voice, or both. The term Generative Pretrained Transformer (GPT) refers to “a language processing model trained on massive data to produce writing that resembles a person’s [language]” (Javaid et al. 2023, 2). “Massive data” is noteworthy: ChatGPT’s development has relied on using “billions of text samples” to train the system to generate responses (Trust, Whalen, & Mouza 2023).
Unlike apps like Google Translate, ChatGPT stores conversation history in a sidebar unless the user deletes it. If the user has additional prompts or questions about previous interactions, the stored history can be referenced for continuity and context.
