Teachers, it's time to make peace with AI

Important: Translated automatically from Spanish by 🌐💬 Aphra 1.0.0

In my previous life as a student, the preparation of monographic works, both individual and group, went through several stages. In the first, we would go to the library (I’m sure this word sends shivers down many people’s spines) hoping to locate among the voluminous volumes some useful paragraph that we could copy manually. A little later, some privileged few of us were lucky enough to have some CDs labeled “Encarta”¹. If the topic wasn’t too complicated, they saved us a trip and the search was now performed by the computer itself. Otherwise, the dynamic was the same: locate the paragraph and copy it by hand. In the last stage, we already had access to the Internet. And with it, the greatest nightmare for teachers who assigned homework: Wikipedia. Of course, they managed to convince us that most of the information it contained was probably false. Because there: “Anyone could write”. It’s curious, because somehow for the first time we were forced to do something different. We knew that teachers would check the Wikipedia article, so it could be our source of information, but we had to slightly change the way of expressing the same thing, we could no longer copy a paragraph literally. Well, at least until El Rincón del Vago² arrived. It seemed that the only solution available to avoid copy/paste was to make us do the work by hand instead of letting us use the computer. “This way we make sure they read the information at least once,” they might have thought. How beautiful and orderly a printed work looked! And how wonderful were the covers made with WordArt³. The fact is that we were never taught to extract and manipulate information, or at least I don’t remember learning it in those stages. It was funny to see the textbooks of classmates, highlighted with “the most important”: Everything, except for determiners, prepositions, and some adverbs.

I can’t help but wonder: What would have happened if Wikipedia had been valued as a resource instead of being pointed out as a natural enemy of teaching? What if we had been sent to analyze an article, its sources, and make a critical comment on its veracity as an assignment? What if a topic with little information had been chosen to complete it with references from other places, in a collaborative and supervised manner? I don’t know, maybe today Spanish wouldn’t be in 8th place in terms of articles, being the fourth most spoken language in the world. Undoubtedly, a missed opportunity.

And here we are, in 2023, about to fall into the same trap again, like good human beings. Starting with the concern about detecting works and essays created with Artificial Intelligence. I’ve even read that teachers have been copying and pasting content from students’ work into ChatGPT to directly ask if it was used to write it. Please don’t do that, because that’s not how it works. The company that created ChatGPT itself made a detection tool available and was forced to withdraw it, because they couldn’t get it to work correctly. There’s another company whose business model is based on detecting texts written with large language models. It uses multiple different approaches simultaneously to give a result and boasts of being “the solution with the fewest false positives”. As teachers, we know that if we want to accuse students of plagiarism, we not only have to be 100% sure, but we also need proof of it. So the best thing, in my view, is to assume from now on that texts, images, videos, and other digital content can be created with Artificial Intelligence. That we cannot and will not be able to detect it with the level of precision we need. From this premise, let’s act accordingly.

When I say act accordingly, it doesn’t mean using it as an excuse to stop thinking of new learning or evaluation formulas and take the easy path. For example, this teacher indicates that he has gone from assigning works that were done with books to going back to doing a traditional exam on those books. Ensuring, moreover, that he doesn’t think that memorization makes sense. Because of course, trying the middle ground of making an exam where they can have those books on hand to consult would be something outside all logic, it wouldn’t occur to anyone with common sense. I’m not one to tell anyone how to do their job. If he found it useful for his students to study the books by heart, fine by me. But don’t sell me that Artificial Intelligence has left you with only one option, and that on top of that you think it’s a solution that harms the teaching-learning process of your subject. Fortunately, there are other examples that allow us to have hope, that invite us to reflect on how we can turn an enemy into our best ally.

It will be nice to see, in this post-ChatGPT era⁴, what new ideas teachers from around the world put on the table. Meanwhile, if we can’t think of anything better, we can do something that’s already been invented. Something that people who teach mathematics know very well: evaluate the process and not the result. Because if someone can use a calculator to solve a calculation that we want them to know how to do manually, we only have to force them to indicate the steps they have followed to reach the solution. And we can do that in many ways, depending on the students, the stage, and the resources. From telling them to search for information at home (or on computers located in the classroom) to write the work during class hours, to more technological solutions, such as using some online word processor that allows reviewing the history of the steps taken to reach the final work. It will also be interesting to give significant weight of the grade to a learning log⁵ that students have to keep. Something that I find very useful to give them feedback and thus allow them to improve their workflow.

And why couldn’t these new technologies be part of that flow? Again, depending on the stage we’re in and the use we give it, it can be very useful. Especially in the later courses. In those where it is assumed that students have learned to write and express themselves correctly. Because let’s not forget that they then enter the working world. A world where money rules. A world where money is exchanged for time. Therefore, a world where any tool that saves time will end up prevailing. Because in the first phase of this revolution we’re beginning, Artificial Intelligence won’t take your job, but a person who knows how to use it to be more productive will.

Obviously, to teach the possibilities and limitations that this technology currently has, teachers must train themselves. On the Internet, we can find enough free resources for this with a little curiosity. Including a guide offered by OpenAI, the company behind ChatGPT. Apart from reading a lot about the subject, it’s also necessary that we put it into practice. We often complain, rightly so, about the number of hours (far above our contract) that we need to dedicate to our work to do it well: session planning, corrections, adapting material to the diversity of our classroom, meetings, project management, tutoring… Some might argue: “Isn’t using AI for grading considered cheating?” And probably that same person has at some point proposed a multiple-choice test that is automatically corrected in the virtual classroom⁶. It’s not about completely delegating any task to Artificial Intelligence, but about using it as an assistant that allows us to do 80% of the management work in 20% of the time. Always in a supervised manner and contributing our experience to the process. Perhaps this technology is the first step to provoke an evolution of teaching. One in which personalized learning will be possible while we improve our work-life balance⁷. Wouldn’t that be wonderful?

Thank you for reading and making it this far.

PS: No AI was harmed in the writing of this article… But it was used to generate the image that illustrates it.