Instructor Profile: Elaine Khoo

Elaine Khoo is working with Contexia, a tool developed by a UTSC team and supported by a LEAF+ grant, that is in the early stages of development. At this time, Contexia is being used to support her students’ engagement with texts. Contexia allows each student to approach the work based on their individual needs as an English Language Learner. The tool will be further developed based on student feedback and interaction.

Q: You use generative AI tools in your courses to support and empower your English Language Learner students. Why do you think generative AI was the right choice?

We are at an exciting time to explore how GenAI can be harnessed strategically to be used in a proactively scalable, personalized, and non-stigmatized manner to meet diverse student needs (e.g. interacting with others, meeting various academic expectations in reading, writing, presenting, etc.). For many multilinguals, English may be their 3rd, 4th or even 5th language, and having tools that can help express the ideas/knowledge they have acquired through their more dominant languages in a fluent manner seems attractive. With the GenAI landscape changing rapidly, students are curious, anxious, fearful, and confused about how much they can use GenAI. As such, using AI in class in ways to support and empower students is the right choice. However, it is important that we focus on using GenAI in the process (as opposed to focusing on the product to be submitted for grading) of learning and language development, and that the work we do in class helps students develop critical AI literacy. They need to understand the good, bad and ugly of GenAI in an experiential way so that they can make informed decisions about using GenAI in active, creative and ethical ways for the process of learning. So far, we try to work with students as partners.

Q: Can you briefly describe how you have used ChatGPT in your courses (e.g., an assessment example or learning activity)?

I have used ChatGPT in three main ways: (a) as an object of exploration and comparison to various GenAI tools I had introduced to the class to explore to support various aspects of language development; (b) as a means of driving home the point that ChatGPT is just a text generator based on its LLM, but it is not to be trusted as a knowledge generator (since its output is based on prediction of what is plausible); (c) as means to provide personalized explanation of meanings of unfamiliar words when students use Contexia, the digital tool my team developed to help student with comprehending their academic texts.

(a) Since ChatGPT is one of many GenAI tools available, I wanted to introduce students to different tools, and engage them in using higher order cognitive skills and their creativity in comparing different tools to meet different language development needs e.g. improving reading speed, becoming more confident holding a conversation in English, preparing for an interview, practice with having small talk, learn about Canadian culture through conversation, etc. So, I organized students into panels for a mini-conference where they presented their findings and their research, and answered questions from their classmates. It was an exhilarating experience for them to experience what it is like to be junior scholars. Everyone provided feedback to the presenters through a MS form, and I compiled all the feedback anonymously and gave each presenter the feedback so that they can make use of it to improve the final essay that they were working on based on the topic of their presentation.

(b) As for demonstrating the ChatGPT is just a language generator, I did a live demonstration with my students on a topic we all agreed on for ChatGPT to produce an output. Then students worked in groups to fact-check, and critique the output. Thus they were able to see first-hand the hallucinations, sentences that seem initially impressive but in reality were merely padding for vague generic ideas, etc. This effort also engaged them in learning about prompt engineering.

(c) Drawing on ChatGPT’s advantage as a LLM, my team has used it to power Contexia, the digital tool developed through the LEAF+ grant. When students are reading their disciplinary texts, they can have Contexia explain the meaning of any unfamiliar words or terminology in the context of the text they are reading online without leaving the page. Beyond single words, Contexia can also explain phrases and sentences in the context of the larger text. This is immensely helpful to students who have many challenging words and phrases on a page they are reading. English language learners say that this support is more helpful to them then the generic meaning of single words provided by the dictionary. Also, using Contexia allows students to reduce their dependence on translating texts into their dominant language in an effort to understand their texts.

Q: Have you noticed a change in student response and/or adaptability as they gain more experience using generative AI tools?

By being able to read their texts, students have found it easier to do their assigned tasks (which include reading many peer reviewed journal articles) and cope with the weekly quizzes. Basically students have less anxiety and fear to deal with academic reading because Contexia reduces the inequity of learning opportunity by being a personalized tutor providing instant support to each student at the point of need. This approach does not feel remedial or stigmatizing to students. It empowers them to carry out their learning more effectively. Students have reported how it now makes it less daunting for them to read their academic texts, and they don’t have to rely on their translators for word by word translation. By being able to read extensively, students are able to expand their vocabulary.