This is a series of articles I will be contributing to the Linked In community about how I am implementing AI as part of my instructional strategy and content creation activities. Part I covers educating my students to use AI properly. Part II will be how an AI service called Cathoven AI works and how it can be used for EFL teaching. Part III will be a deeper dive into using AI to discover better lesson plans, materials, and subject matter for the classroom
I am an older teacher. I like certain traditions. I like planning things for myself. I like watching students finally ‘get’ the gist of what’s going on in a language that isn’t their own. I constantly revisit what and how I teach to make sure it’s all relevant…personal, local, and meaningful. That’s what I strive for.
I am an older teacher who holds a grudge against many things: Cheaters, plagiarists, inattentiveness, cellphone gazers, internet wanderers and all those other things students do that can transform classrooms into unhappy places. Until recently, AI was also on my grudge list.
My grudge was not unfounded. I started catching my students using ChatGPT to answer a writing question on my exams. What started with one student AI user at the beginning of last semester grew into entire classrooms of students in just a few weeks time. How did I catch students who used AI? It was easy. The grammar that EFL students are able to use stood in stark contrast to the crystal clear, sparkling prose of the AI generated work. That was first. The second clue was that the AI generated content failed to answer the question that was posed. It did do a remarkable job of evading the task and suggested several logical avenues to resolve the problem in the essay question.
My first reaction was to stop using my usual tests and revert to standard TOEFL or TOEIC type tests with multiple choice questions and listening and reading comprehension tasks. I even abandoned Moodle and went back to using paper assessments. This, of course, increased the amount of time I spent grading. It was during one of my marathon grading sessions that the realization hit me: I needed to rethink my views about AI.
I am an older teacher. I dislike AI, but I also dislike not coming to grips with the fact that AI is becoming more prevalent. So, I have decided to allow my students to use it on the next exam. The caveat here is that students will have to learn how to use it correctly to get an acceptable answer that they can use as part of their assessment.
My language assessments are in the form of a jigsaw puzzle test over a topic like YouTube or TikTok influencers. That is, Student A gets a video, audio, or reading about an aspect of the main topic, and their partner, Student B, gets material that is related to the topic, but contrasts or adds to the information that their partner has. Students can work on the answers together, use the internet for more information (if they have the initiative to do it…most don’t) and are even allowed to talk to ‘experts’ they may know in the community (see initiative comment). Their final question requires them to combine the things they have learned from their respective test material. Here’s an example:
Well, you and your partner have just gotten an email from a good, old, mutual friend, Tonio Ñoño. Tonio has a great idea. He wants to become an influencer. He has been making videos about racing his car illegally, and winning, a lot. His videos have several thousand views, and he is often in trouble with the police, his family and his work. Tonio hasn’t always been very honest, or hardworking, but he’s a friend. So, he tells you he wants to do a vlog called Ñoño Rápido, and he thinks it´s a great idea! He wants to borrow your money and your advice to get started. Thankfully, you and your partner have just learned all about influencers in your English class, and together you decide to give him advice, but not money, of course ( he will have to make that himself). You could tell him how he could make some money, though.Together with your partner, decide what to tell Tonio. You can tell him anything you want as long as you show you have based your evidence on the things you learned in this collaborative learning opportunity and the information you learned by talking about influencers in class and on your own. Structure things you might want to include would be modals of obligation (could, would, should, must, may, etc.) and the first two forms of the conditionals.
I discovered that my students were using AI with the answers to this question last year. Where the students messed up was in the details. Chat GPT didn’t have access to the materials the students read, so it couldn’t give an answer based on what the students learned. As a result, the AI was surprisingly creative in constructing an answer based on the information it did have, and it did so using an extensive list of modals and conditional forms.
So, what to do? I am going to allow my students to do this exam as an in-class assignment and ask them to use AI to answer the essay question. Then, the students will evaluate for themselves all the responses. I will even throw in examples from last semester. There will be only one top grade, so they will have to choose the best one. As the AI responses will be nearly identical this will become a very difficult task. Given that students will likely copy and paste their answer without closely reading what the AI produced, it will give us all a chance to examine the responses in a more critical light.
This will hopefully lead to a discussion about how to best incorporate AI as a tool for such a task. The desired outcome will reflect that students can use AI as a tool, but not the final product, as their input in the final written production is essential.
Additionally, I will press the students to create a set of evaluation standards that will be applied. I am an older teacher, but I do like it when students participate in their own assessment. Our spring break finishes next week, so I will implement this plan very soon. How will it turn out? I will let you know!