Recently I was remembering the famous painter Bob Ross’s program: The Joy of Painting. For 30 to 40 minutes, Bob taught us step by step how to turn a blank canvas into a work of art. Along the way he reminded us that it’s okay to make mistakes and that anyone, if they wanted to, could follow his steps and create the same artwork. This American program was broadcast on public television channels in that country, and on Reddit there are testimonials from people who did manage to recreate these paintings, proving that Ross was right when he said “imperfection is what makes something beautiful, that’s what makes it different and unique from everything else.”
I’ve rarely tried to paint, but I related this feeling to programming. It took me three years of college to find my taste for it, but as time went on I began to find topics that passionate me and to appreciate other people’s code: seeing the way someone solved something particular after spending a couple of hours reading code in their repository, the difference between libraries that solve the same problem and, of course, seeing something I had created after hours being used by other people.
At the same time this was happening, I was given the opportunity to be a mentor in a high school robotics club. This consisted, mostly, of teaching the fundamentals of programming with Java and using the libraries the robot needed. That was my first experience in front of a group, I had no idea how to teach a class so the first sessions were just presenting a topic, just like in university. The problem is that due to their age, and the time I had to teach the class, I couldn’t maintain their interest and those who paid attention to me didn’t quite understand the topic. Looking for a way to communicate with them, to speak the same language, I realized that a love for anime was the common denominator among all of us in that classroom. So I began to change technical terms or generic exercises for examples related to anime: the exercise to practice conditionals went from being a grading system to a simplified dating simulator. Objects went from being cars or animals to character archetypes in anime. With this I managed to connect with them and they managed to learn what I wanted to teach them. This planted the seed of wanting to teach.
Several years after graduating with my bachelor’s degree, I had the opportunity to teach a subject for the same program I had graduated from, a former professor (now colleague) of mine had recommended me for it. I gladly accepted and began teaching in the same way I had done with the robotics club: as best I could. With this group I found the common language differently, giving them the opportunity to express themselves uncensored in front of me and offering them my advice in the academic and work environment. That semester I was among the ten best-rated professors in the Engineering Coordination, so this year I started my master’s degree in education, with a focus on online education. I was sure that, while it wouldn’t be my full-time job, I had been wanting to teach for years. Now we’re mid-semester with the second group I’m teaching this subject to but things couldn’t be more different.
In one year, the artificial intelligence landscape has advanced a lot. A whole lot. Both technically, with more models, larger and more tools to use them, and in people’s confidence with them. A year ago, at work, I used to use some LLM to get me out of some doubt, solve an error or create a small block of code. Now, the same company gives us tools so that, with the right ticket, we just give the instruction to the chatbot and supervise it while it solves the ticket in its entirety. Outside of work, I implemented a script to update the “latest reads” section of the blog according to my Raindrop account. I went from idea to execution in an hour, but I didn’t feel any sense of accomplishment for having done it.
My students now use some form of generative artificial intelligence in all the assignments and projects in the class. The exercises (designed to make them reflect, research a topic or learn to use a tool) go: from the school platform to a chatbot and back to the platform. You can rarely see that the students did most of the work, most just read the response that the models generated for them. Some don’t read it, they just copy, paste and, if they have to present it, they read the slides.
At work, used the right way, artificial intelligence is more efficient for me and for the company. It gives me a better work-life balance. In the same way, why wouldn’t my students use it in class? They turn in what I ask for, they solve the problems and before the exam they study enough to pass. The only thing left out of the equation is the joy I felt for teaching. This semester, I’ve rarely felt it, and, on those occasions, it was with activities precisely designed so that students couldn’t use any LLM. But those activities are difficult to conceptualize, especially in this area.
I don’t know if it’s my failure as a teacher, for not trying hard enough. I don’t know if it’s lack of knowledge. In the master’s program, I’ve understood that it’s necessary to learn how to teach. I don’t know if my students shouldn’t use artificial intelligence in school, since it’s something they’re going to have to use in the workplace or they’ll be at a disadvantage. I don’t know if it’s the system, which has to be reformed in the face of a paradigm shift. The only thing I know is that I no longer feel the joy for programming or for teaching. There are no longer, as Bob Ross said, mistakes that are just happy accidents.
Originally published on El telar blog