This is part of an ongoing GenAI op-ed series, Arts Perspectives on AI (Artificial Intelligence), that features student and faculty voices from the UBC Faculty of Arts community.
This image was generated with the assistance of Artificial Intelligence.
By Ritwik Bhattacharjee, Communications Specialist and PhD candidate in Interdisciplinary Studies
Human languages are complex and ever-evolving. Learning to speak a language is as much a matter of grammatical and linguistic correctness as it is one of immersing oneself in the social and cultural practices that make the language ‘living’.
More people are turning to AI for help with learning new languages and celebrating its unquestionable strengths – such as providing immediate feedback, creating language chatbots for quick conversations, and crafting routinized yet customized lesson repetitions. However, what has also come to the forefront is AI’s lack of contextual understanding, that is its lack of genuine comprehension of purely human interactions. Such drawbacks are proving more problematic as classroom uses of AI continue to be on the rise. Should we stop using AI altogether to avoid transferring its socio-cultural blindness to students eager to learn new languages? Or, can we train AI to capture the subtleties of human interaction?
Dr. Muhammad Abdul-Mageed (he/him)
Canada Research Chair in Natural Language Processing and Machine Learning, Associate Professor in the School of Information and Department of Linguistics, and the Department of Computer Science (courtesy)
Dr. Abdul-Mageed shares his thoughts on these pressing issues. He argues that only by focusing on the social embeddedness of AI, rather than merely using it as a technological tool, can we help attune AI to human needs. Questioning the lack of equity in current multilingual models, Dr. Abdul-Mageed highlights how training students to build robust language models that are aware of their larger societal implications should be at the heart of conversations about using AI to teach languages.
How does using AI to learn new languages differ from traditional methods?
AI-enabled language learning offers a distinct departure from traditional methods by providing more flexible, personalized, and adaptive experiences. Unlike conventional textbooks or standard course materials, AI can deliver richly multimodal content and tailor exercises to an individual’s progress in real time. This approach not only offers immediate feedback on grammar, pronunciation, and vocabulary but also extends to sophisticated writing assistance and, increasingly, cultural guidance.
Such responsiveness promotes more efficient study, enabling learners to engage interactively within simulated real-world contexts at their convenience. As AI technology continues to evolve, it also introduces learners to a wide range of dialects, accents, and cultural contexts, effectively bridging gaps that traditional methods may fail to address.
What are the major risks and opportunities that AI poses to language learning?
While AI offers accessibility and convenience, it also brings challenges. Automated tools may not fully capture the subtleties of culture and socio-pragmatics (such as emotion and humor) inherent in human communication. Over-dependence on technology also risks diminishing social interaction, and AI systems trained on biased data can introduce inaccuracies or reinforce stereotypes. Privacy concerns may also arise when users interact with AI systems that collect personal data to improve learning. However, when integrated thoughtfully, AI can truly enhance language acquisition.
“Unlike conventional textbooks or standard course materials, AI can deliver richly multimodal content and tailor exercises to an individual’s progress in real-time.”
Tell us about your approach to teaching students about AI practices.
I design my courses to foster deep reflection on the cultural, social, and ethical dimensions of AI, nudging students to ask who benefits from these technologies, who might be excluded, and what unintended consequences might arise. This involves examining biases baked into training data, grappling with fairness and accountability in machine learning systems, and considering the privacy implications of data collection, as well as learning fundamental concepts and algorithms.
I also believe in guest lectures and interdisciplinary collaborations that further broaden students’ perspectives. This comes naturally to me as I am cross-appointed between Linguistics, the School of Information, and Computer Science. I approach teaching AI with the aim of training well-rounded experts who are as skilled at building robust models as they are at critiquing their broader societal implications.
What can we learn from your research on AI practices?
One key lesson from my research is the necessity of viewing AI as a socially embedded technology rather than as an isolated technical tool. The performance and impact of AI systems, such as large language models, are profoundly shaped by the data they are trained on and the communities they are designed to serve. My research demonstrates that multilingual models, when designed with equity in mind, can foster greater understanding and support a wide range of applications, including cultural heritage and archiving, education, health, business, and safe social networking.
“I approach teaching AI with the aim of training well-rounded experts who are as skilled at building robust models as they are at critiquing their broader societal implications... One key lesson from my research is the necessity of viewing AI as a socially embedded technology rather than as an isolated technical tool.”
What do you believe are the biggest AI-related opportunities and/or challenges facing faculty and students?
AI presents transformative opportunities to enhance research, creativity, and productivity. For students, AI-powered tools offer personalized and accessible education tailored to their individual needs, helping to democratize learning by providing resources that were once unavailable or cost-prohibitive, particularly for underrepresented communities. Faculty, too, can harness AI to streamline administrative duties, enhance the quality and reach of their teaching, and push the boundaries of research. However, these opportunities come with significant challenges, such as navigating the ethical implications of AI usage, especially surrounding academic integrity and data privacy.
Is there anything else you would like to comment on that hasn’t been addressed?
One additional point I would like to emphasize is the irreplaceable role of humans in the development and deployment of AI systems. In education, for example, teachers play a crucial role in providing cultural context, fostering critical thinking, and guiding learners through complex social interactions. Moreover, the environmental impact of AI, especially the significant energy demands of training large-scale models, must be addressed through sustainable development practices.
“AI presents transformative opportunities to enhance research, creativity, and productivity.”
Dr. Abdul-Mageed is grateful to live and work in the UBC Vancouver-Point Grey campus on the traditional, ancestral, unceded territory of the xʷməθkʷəy̓əm (Musqueam) Peoples.
About the featured image
The featured image was generated with the assistance of Openart.ai which is an AI-powered image generator.
Dr. Abdul-Mageed provided the prompt: “A high-resolution scene of people learning a new language with the help of AI technology. Multiple human faces of different ages and ethnicities engaged in communication, smiling and focused. Computer screens in the background displaying text, linguistic symbols, and futuristic AI graphics. Subtle digital or holographic elements floating around to emphasize artificial intelligence. Warm, inviting atmosphere, suggesting collaboration and innovation. Hyperrealistic photography style, balanced lighting, vibrant colors.”
About the author
Ritwik Bhattacharjee is a PhD candidate in the Interdisciplinary Studies Graduate Program and a Communications Specialist for the Faculty of Arts. His interdisciplinary project works towards dismantling settler colonialism by investigating the role of socially unconscious background assumptions that are resistant to transformative change.
