To me, it doesn’t seem crazy to say that regular AI use is bad for the brain. Yet, we’ve adopted the technology so quickly that even raising this concern is met with “the cat’s out of the bag.” Well, now that AI (of the generative variety) is seemingly ubiquitous, I’m starting to see the effects in my day-to-day interactions with students.
Table of Contents
- Just Prompt Harder, Bro
- No One Wants to Work With the Helpless
- When Prompting Doesn’t Work
- The Wall-E Future
Just Prompt Harder, Bro
Forgive me if you heard this story already, but very early this semester I had a student run into some issues in lab. They were unable to get my repo working in VS Code, and they wanted to see if I could fix the issue.
When I got to them, they were getting an error I hadn’t seen before—something about a version mismatch. Having just updated our API, I thought maybe they just needed to install a newer version of Java and we’d be good. Yet, when I checked their current Java version, it was already the latest.
At that moment I was stumped, but before I could even begin troubleshooting, they said “why don’t we ask AI?” I was extremely confused by this because we hadn’t even tried to do anything, yet they were ready to throw in the towel.
As I brushed off his comment, I noticed that their repo was incredibly broken. Their settings file contained several new entries that I never added, and they even managed to change a few global settings to my amazement.
Apparently, they had been using AI to “advise” them on how to address the error. As a result, they managed to put their repo in an almost unrecoverable state.
No One Wants to Work With the Helpless
While this was my first interaction with the student, it certainly would not be my last. For example, during team formation in the first week, they claimed to have found a team, but their “teammate” told me in private that they didn’t want anything to do with them.
Later, things started to click for me. The reason no one wanted to work with that student was because of their obsession with AI (or perhaps their learned helplessness). I found that out in a conversation with one of my peers. Apparently, they know each other, and the peer told me that the student subscribes to Claude Max, which runs them over $100 a month.
While paying that much for a subscription is diabolical, I learned that it paid for itself. The student had apparently won several hackathons using just Claude Max, enough to pay for the subscription. And, honestly, I think that says more about the state of hackathons than “AI.”
I heard from that same grader that their basic strategy was to bring two monitors: one for prompting and another for gaming. I guess when you leave all of the software development to a bot, you can enjoy yourself during the downtime.
When Prompting Doesn’t Work
To be honest, I don’t think anything you’ve read up to this point is worthy of an article, but something interesting happened to me this morning. Today, I opened up my notifications to see that same student ask the following:
Does anyone know when our software two final exam is like the exact day and time?
I don’t know what bothers me the most about this post. The fact that this information has been posted on our learning management system since before the semester even started? The fact that the vast majority of finals are determined by the timing of the first class of the semester (and therefore can be determined by referencing a lookup table)? Or, the fact that their brain is so rotted from prompting that they think they can treat people like chat bots?
Like, it’s definitely that last point, right? Because the information for the exam schedule is hidden from the internet; they can’t simply prompt Claude Max for the answer. So, when your primary and only source of information can’t answer your question, what do you do? You lack the ability to solve problems.
I used to always get annoyed by folks who say things like “read the fucking manual” or “just google it.” I always felt like that was extremely isolating and probably turned a lot of folks away from the field.
At the same time, however, part of the social contract for coming to others for help is to demonstrate that you have tried to solve the problem yourself. Otherwise, it looks like you’re trying to get someone else to do your work. Like, this entire dynamic is based on the good faith assumption that you put in some effort.
And look, I’ve been embarrassed plenty of times for asking “dumb” questions. Most recently, I got roasted by one of my Japanese teachers for not knowing a concept that was defined in like the second chapter of the book. I think they could have been kinder, but now I know I need to come better prepared.
In contrast, with folks like this student, there was never an attempt. Prompting does not signal an attempt to try to me. It signals giving up before even starting. Like I know this has been said to death, but watching someone (or some “thing”) do the work does not constitute doing the work.
The Wall-E Future
I don’t really know what the point of this article is. After all, I’ve ranted about AI to death on this site, so what value could another anecdote have? For me, I guess it’s catharsis. It just feels good to get some of these experiences off my chest.
That said, it’s really hard not to have like a Lisan al-Gaib moment every time I see a student pop open ChatGPT. It’s like I can see directly into a future where no one has skills, interests, beliefs, thoughts, or feelings. Everyone is just strapped into a NerveGear chanting “grok, is this real?” at AI-generated assets until its time to return to the data center labor camps.
Dystopia aside, I’ll close with a wonderful rant from another student about using LLMs to code. It’s a bit elitist for me, but they basically said: “if you’re letting a machine write code for you, you clearly don’t like software development. If all you care about is building something, why not do blue collar work?”
I found this fascinating because it perfectly highlights the current mindset. AI exists to produce stuff. That’s it. If you value the process over the outcome, then you probably won’t use it. But, everyone else will continue to “build” stuff.
Once again, thanks for reading! I’m crazy busy at the moment, so I haven’t had a chance to write anything deeper than my usual rants. Hopefully, these types of articles are still interesting to you.
That said, I suspect a lot of my work will be similar for a few weeks. Right now, it’s rapidly nearing the end of another semester. Next week is my last week of classes before finals week. I have to quickly work through grades, which will go out early May.
After that, I immediately hop one a plane to Japan again. As you can imagine, this transition period is particularly painful. I’m setting up the learning management system and prepping for the study abroad class while trying to wrap up a different class. It’s a lot to balance, but I’m excited for it to be summer.
Anyway, if you liked this piece, there are plenty more AI hater pieces like it:
- Why I Rebel Against the Use of Generative AI
- I Hate That Student Feedback Is Now Reviewed by Machine Learning
- No, Generative AI Is Not Just Another Innovation
Likewise, it would help me a lot if you checked out my list of ways to support the site. My accountant would love it if the site didn’t bleed money every year.
Of course, no pressure! I’ll see you next time.
P.S. I have another short story I would like to share, that I’ll just sneak here. While recently facilitating a think-pair-share activity, I overheard a student ask their peer if they were “actually doing the activity,” as if it was almost absurd. The response was something to the effect of “yes, I’m actually trying to learn.” Five years ago, I would have found this interaction funny. Today, it’s just sad. After all, the student rejecting the activity literally argued that there’s no point in learning if a chat bot can do it for you. Thankfully, the conversation continued with “(Student A) I still want to learn. (Student B) But, isn’t that hard? (Student A) Yeah, it sucks!” And you know what, learning does suck. I get why people don’t want to do it. Some days I just want to toss out my Anki deck and give up Japanese, but there is just something deeply fulfilling about seeing myself grow and improve. There’s nothing else like it.
P.P.S. A couple notes from this article. The student who talked about blue collar jobs also ranted that they noticed a lot of peers using ChatGPT to get through their first programming class. This bothered them immensely, in almost the same way as the title of this piece: quitting before even giving yourself a chance to learn. I recall some of my peers making similar comments in office hours. It seems like a lot of folks are bothered by this.
P.P.P.S. I’ve also been thinking about how prompting is such a silly way of navigating the world. It’s so imprecise. It requires almost mystical levels of experimentation. Do I praise the bot? Do I yell at it? How should I phrase this question exactly? And, it all changes every time they push an update. That’s why I used the phrase “just prompt harder, bro.” It reminds me of a fun conversation you can have around how superman’s flight works. Does he just think really hard about it? I’ve certainly tried that a few times. Like, maybe if I focus on a spoon hard enough, I’ll be able to move it with my mind. This is the exact level of absurdity I experience when people are like “you’re just not using ChatGPT right.” Brother, what skills are there to master? It’s a chat bot.
P.P.P.P.S. I have to sneak this one right at the end because I don’t expect anyone to read this far, and I’m a little petty. The student much of this article is about has failed both of their midterms. Like, is that not evidence enough that LLMs are not helping folks learn? My midterms are comically easy (because I hate them)—we’re talking medians in the mid-80s. I mean I expected more students to struggle once I switched back to paper exams, but averages seem to be about the same. Yet, the person who might have AI psychosis is incapable of passing them.
Recent Blog Posts
Duolingo is a fun little app for picking up the basics of a new language, but you will never truly learn that language. Trust me! I've tried.
Hey, what can I say? Haven't I ranted about efficiency enough? Would you prefer I were more succinct or should I say "efficient" in my writing?