ChatGPT as a Collaborative Thought Partner: Opportunities and Challenges for Multilingual Writers

The rise of ChatGPT is transforming writing, offering new tools to increase efficiency while also exposing alarming inequities. For multilingual learners, these tools hold promise but also perpetuate systemic challenges. Biases in AI systems, such as disproportionately flagging multilingual writing as AI-generated, unfairly penalize these students and undermine their efforts. From AI text detectors that disproportionately flag multilingual writing as machine-generated to the reinforcement of cultural and linguistic biases, these technologies can create as many barriers as they dismantle. Sharin Jacob, researcher for computational thinking pathways at Digital Promise, examines the opportunities and challenges of generative artificial intelligence (GenAI) tools such as ChatGPT, balancing their potential for collaboration, reflection, and empowerment with a critical awareness of the biases and inequities that multilingual writers face in an AI-driven world.

Hidden Bias in AI Tools

AI text detectors, widely used to identify machine-generated writing, often work against multilingual writers. For example, biases in AI-generated text detectors are a critical concern, with detectors mistakenly labeling English essays written by multilingual students as AI-generated at alarming rates, with over 61% detection rates in some studies. Additionally, generative AI tools often perpetuate stereotypes and biases embedded in their training data, further marginalizing students’ writing and expression by reinforcing and exacerbating linguistic and cultural inequities. . Multilingual students deserve tools that support, not hinder, their learning journeys. Without addressing these systemic issues, AI risks becoming another barrier rather than a bridge to equitable education.

ChatGPT as a Writing Resource: The Power of Collaboration

Researchers from Digital Promise and the University of California, Irvine (UC Irvine) sought to understand whether ChatGPT could be used as a collaborative writing tool in a manner that preserved authorial voice and integrity. A recent case study titled, “Emergent AI-assisted discourse: A case study of a second language writer authoring with ChatGPT,” written by Sharin Jacob (Digital Promise), Tamara Tate (UC Irvine), and Mark Warschauer (UC Irvine), found that when used in concert with collaborators such as colleagues, field experts, editors, and peers, the tool can facilitate academic writing in a way that preserves authorial voice. This was the case for Kailing, a multilingual doctoral student featured in the case study, who was tasked with developing an academic research proposal. Kailing utilized ChatGPT to brainstorm ideas and generate initial drafts but then refined those outputs by consulting academic literature and seeking feedback from her peers to ensure accuracy and depth. She critically evaluated ChatGPT’s responses, cross-referenced citations for validity, and adjusted the text to align with her personal writing style. By treating ChatGPT as a collaborative “thought partner” rather than a substitute, she retained control over the writing process and ensured her work reflected her unique intellectual contributions.

Kailing’s writing process can be understood through the lens of activity theory, which highlights the dynamic interactions between tools, individuals, and their broader sociocultural context. Within this framework, ChatGPT acted as a mediating tool that supported her goal of producing an academic research proposal. As a multilingual student, Kailing leveraged the AI tool not as a passive instrument but as an active collaborator, using it as a tool to brainstorm, refine, and produce text while engaging with academic norms, peer feedback, and disciplinary expectations. This interaction was shaped by the rules of academic writing, the community of scholars and colleagues with whom she collaborated, and the division of labor involved in refining her drafts. By critically evaluating ChatGPT’s outputs and integrating them with her expertise, Kailing demonstrated how generative AI can mediate the writing process, enhancing cognitive and linguistic efforts without undermining the writer’s agency or voice.

From Enthusiasm to Critical Reflection

When Kailing first began using ChatGPT, she was impressed by its capabilities. The tool provided quick responses, polished language, and efficient text generation. However, as she continued to use it, she noticed its flaws: repetitive patterns, shallow ideas, and a lack of nuanced thinking. These limitations led her to evolve her approach, using ChatGPT for specific tasks such as paraphrasing or diversifying sentence structures while relying on human collaborators for deeper intellectual engagement. This shift underscores the importance of approaching generative AI tools such as ChatGPT critically, rather than uncritically adopting them as the latest technological solution. While these tools can enhance efficiency, they cannot replace the depth of human insight, creativity, and ethical considerations that are central to academic writing. By thoughtfully engaging with ChatGPT and acknowledging its limitations, multilingual writers like Kailing not only refine their work but also ensure that their unique voice and intellectual contributions remain intact.

Moving Forward: Equity and Empowerment in AI Writing

Kailing’s journey underscores the potential of ChatGPT to level the playing field for multilingual writers if used critically and equitably. To truly leverage this potential, we must:

1. Address bias: Developers must refine AI systems to eliminate biases against multilingual students and ensure fair treatment for all users.
2. Promote critical engagement: Educators should teach students how to use AI tools critically, focusing on preserving their authorial voice and ensuring academic integrity.
3. Foster collaboration: Institutions can create spaces where students, educators, and researchers collaborate to explore the best practices for integrating AI into writing.

Generative AI tools are reshaping the landscape of writing. For multilingual writers, they hold the promise of facilitating linguistic expression. But this promise can only be realized if we approach these tools with both caution and curiosity, ensuring they serve as allies in the pursuit of equitable education. Future research should focus on leveraging AI without blindly relying on detectors as well as exploring how training data can be curated to mitigate bias. Additionally, studies should examine how multilingual students can be empowered to navigate these challenges while maximizing the benefits of GenAI tools. Using activity theory as a framework, GenAI tools can be understood as mediating instruments that support writing processes within broader sociocultural contexts, academic norms, and collaborative interactions. By balancing equity considerations with supportive frameworks, we can ensure that AI tools not only reflect equitable practices but also actively support the diverse needs of multilingual writers.

Learn more about AI literacy and read our AI Literacy paper. Sign up for our newsletter and follow Digital Promise on LinkedIn, X, Facebook, and Instagram to stay updated on computational thinking.