I’m going over a bunch of papers talking about using AI in classrooms and education in general. In the background I’m listening to The Vergecast on AI in music and it got me thinking… many times I’ve seen or heard that the current expression of AI can be seen as an assistant or co-pilot (I can’t believe I’m saying MS has had the best idea because I really like that description – ok, younger me wouldn’t have current me can say that and not care). So the systems are not doing or creating on their own. They are simply taking some of the heavy lifting around iterating from an increasingly wide array of tasks performed by the student and the instructor.
Getting ready to be a lab instructor for the first time in decades, I’ve been thinking about how I would talk to students who are in my sections about AI. The course is about data visualization so suggesting AI be used for dataset generation or caption generation are easy. What struck me is that half of that list I could do way back in my high school days on my Casio 9300 or if I had to slum it, on the TI-83. Creating data sets out of the blue are nothing new. The caption part is novel. It might be a stretch for some, but this is how I’m seeing many of these AI models as mere calulators for text. Very few people had issues with numbers that were generated randomly based on some parameters; but now it seems there is much pearl clutching about text especially and music and visuals to a lesser degree. I wonder why.
My take on why is that the great majority of calculator users are not numerate enough to use calculators in creative ways or find patterns and be drawn into dark holes such as large primes or solving pi to just one more decimal. Teachers were worried about calculators in the classroom and exams. When they started to see that the GIGO principle held they may have relaxed. It allowed teachers to assess and explore a greater space of numeracy with the calculator doing some of the lift for the students. Students still had to be able to interpret the figures on the screen, but that conversation seems to have created a wall. On one side are numbers that mean nothing unless the student on the other side knows enough to find them. Generative AI, looking at those models that generate language don’t have this wall.
With language being on both sides of the divide, students who are not otherwise skilled or literate enough could get “sucked in” and get trapped in a cycle of GIGO. This might be what non numeracy oriented teachers are afraid of. Now it’s not that “over there” stuff of “math things”. Now the little box can create within their lingua franca and they have to up their game. In this regard AI takes the classic academic foul of plagerism to a completely new level. It is hard to plagerise something that never existed. It is hard to develop a voice and a style within a student. But with tools and media homogenizing both how does a teacher know what is the human voice and what is the machine?
So how can a teacher or educator of any stripe get around this issue? As I started this blurb… use it as an assistant. Educators need to play with the machine and get to know what it is. Photographic arts have long played with the idea that photos can be faked and that is part of the way things are so I have a very good feeling what I’m saying is not impossible.
When creating assignments come up with not only the classic “things that can’t be googled” style prompts, but now assume that every student will be working with not only their peers but an increasingly competent (though occassionally halucinating) coach. Assessing the work of the student should also have an emphasis on reflection or “show the work” so regardless how much the AI contributed the student input can be highlighted. It worked in math classes for ages, it is part of the artful process in photograpy. So I feel that educators should be able to accept the use this new and incredibly quickly changing and dispursing tool work everywhere by encourging their students share what and how they wound up at their final result.
For full transperancy, I’m using Co-Pilot to edit this post. It’s frustrating because it makes things sound really bland. Even after repeated drafts, I have not accepted any changes.
TLDR – have students show their work so if nothing else you as an instructor can keep up with what and how they are using the tools that are out there. “Know thyself, know thy enemy” ala Sun Tzu.
Leave a Reply