Apologies ahead of time for another generative AI rant, but I’m trying to get these out of my system before my life revolves around another baby. In the meantime, you’re stuck learning about how generative AI in education is pay-to-lose.
Table of Contents
Pay-to-Win vs. Pay-to-Lose
If you’re not much of a gamer, you might not be familiar with the term “pay-to-win.” It basically describes a situation where a video game offers players an advantage for making in-game purchases. This monetization model is so hated in the gaming community that I don’t think I’ve personally played a single game that leveraged it, but apparently it’s common in mobile games (e.g., Clash of Clans).
That said, I’ve played plenty of games that employ “pay-to-lose,” a sort of tongue-in-cheek reversal of that term. Certainly, I’ve played games where character skins made me easier to see (e.g., Fortnite) and gun skins made it harder to see/aim (e.g., Valorant).
Both of these terms are really interesting to me because I don’t think they have to only describe the design of video games. In fact, once you start to see pay-to-win in gaming, you start to see it in real life too.
An example that comes to mind is sports. Kids with rich parents have a huge leg up. Their parents can afford equipment, coaches, and travel. Even “natural talent,” if such a thing exists, can’t compete with all that.
A similar argument can be made about education. If a family can afford to buy a house in the best public school district or send their kids to the best private school, those kids are already at a distinct advantage over other students.
In other words, a good portion of life is pay-to-win.
Life Is Mostly Pay-to-Win
Yet, what I’m finding really funny is that maybe we’ve reached the limits of pay-to-win as a society. After all, the biggest pay-to-win scheme right now is generative AI, but are the folks using it really winning?
Take education as an example. Sure, students with generous parents can afford to pay $200/month for Claude Max 20x—the best model Anthropic currently offers—but what advantage does that give them? I mean this genuinely. What value does this possibly add to their kid’s life?
On one hand, they’re able to complete assignments very quickly. That seems like a major advantage over the poors. Why fumble around with the Gemini model provided by the university, when they can afford to dump the entirety of each class on Canvas into Claude Fable 5?
On the other hand, here’s what’s lost:
- The student never learns anything because all friction is smoothed over by a chat bot.
- The student never engages in problem solving because every solution is just a prompt away.
- The student never improves and perhaps even worsens their social skills because every social interaction is replaced with a sequence of prompts.
- The student develops psychosis from using the AI for anything unrelated to “learning,” such as therapy or companionship.
Sure, there are a lot of silly counterarguments like, “well, it’s all about how you use the tool.” I’ve seen plenty of folks claim they can use chat bots to build study guides (even if there’s value in creating one yourself) and quizzes, but for every “good” use case there are about 1,000 bad ones.
After all, what seems more likely: a student doing their assigned reading and having the bot give them questions on it OR a student reading the bot’s summary of the reading? In my experience, students (of which I include myself in the sample) generally have bad study habits. What makes you think AI is going to change anything?
For the record, having come from an engineering education background, I don’t feel good about making the previous argument. After all, much of the work in engineering education revolves around making research more asset-based as opposed to deficit-based. In other words, we should focus more on what students bring to the table rather than what’s “wrong” with them.
However, I find both framings troubling because I feel like I should be able to critique the systems that lead to student issues. For example, my dissertation looked at gaps in values between students and their institutions. In some cases, I did not side with the students in my analysis (e.g., students don’t value reading, but I believe they should). Is it bad to point this out? If it’s a symptom of a bigger problem, I believe you must point it out.
In this case, while I have no empirical evidence to prove it, I believe most students have bad study habits. If AI makes it much easier to bypass the one thing that gets students to study (i.e., formative assessments) AND if AI makes it trivial to produce the few artifacts that students might create themselves (e.g., study guides, cheat sheets, etc.), then I would argue AI is not going to have the positive outcomes for students that you’re being led to believe.
Not to mention that I think it’s okay for students to have bad study habits. In my experience, a lot of high achieving students never have to learn how to study until they’re sufficiently challenged. At that point, you’d be doing a disservice by depriving them of their chance to learn how to learn. If you hand them a chat bot the moment they get stuck, you might as well be tossing their potential in the trash. Like imagine if you were coaching a baseball pitcher that started to plateau, and you said, “it’s fine, you can do steroids”—or better yet, just replace them with a pitching machine. Brother, you didn’t even give them a chance to work through it. I have a feeling this is what’s happening to some of our best students.
Therefore, generative AI hardly looks like pay-to-win in education. Students with access are very much self-sabotaging, and their parents (and in same cases, their universities) are footing the bill. If college was a bad investment before, using an LLM to bypass whatever value remains is an odd choice. But hey, your boss wants you to be good at using AI, so who’s the real winner?
Side Note: you might know I’m always annoyed when folks compare generative AI to the calculator. More recently, I’ve sort of accepted the comparison, only because I think they both serve a similar role in short-circuiting learning by fostering a fixed mindset. After all, how many adults do you know that say something like, “math just isn’t my thing?” And, you think it was a good idea to enable that belief by handing them a calculator?
I mean, I literally just saw a paper that referenced this 1998 study where participants were asked to estimate calculations before entering them into a calculator. Despite the calculator being programmed to give increasingly incorrect answers, the participants often trusted the calculator instead. Isn’t that deeply troubling in the age of AI? People are not going to question the hallucination machine, and they’re going to start saying stupid things like, “reading isn’t my thing.”
Work Harder, Not “Smarter”
One of the reasons I think we were so quick to adopt generative AI—ignoring the vast network of capital pushing for it—is that we’re obsessed with the sentiment expressed in the phrase: “work smarter, not harder.” Again, even this slogan is productivity slop, but I get the appeal. Why work hard when there are easier ways to do a task? I mean, my catch phrase for the longest time was literally, “there has to be a better way.”
But as I’ve gotten older, wisdom tells me that often the “slow” way or the “hard” way is the “right” way. There are no shortcuts.
To me, this outright refusal to work hard is concerning. Hard work is how expertise is developed, relationships are fostered, and community is built. I refuse to contribute to the erosion of hard work for fear that it might follow the trend of anti-intellectualism. Y’all really trying to make WALL-E a documentary.
Hi, all! I’m taking the wrap up portion of the article under this horizontal rule because I really liked where this piece ended naturally. As always, I have some relevant articles I’d like to point you toward if you’re looking for more to read:
- Not All Code Completion Is Generative AI
- Why I Rebel Against the Use of Generative AI
- Generative AI Makes It Feel Bad to Be an Educator
And, if you’re looking to support my work, though I know times are tough, you absolutely can by heading over to my list of ways to grow the site. Otherwise, thanks again for stopping by!
Recent Teach Posts
For all that hate that handwritten code gets, I'm starting to think there is some merit in doing so.
Well, it's been almost two weeks since I've returned from Japan, so I figured it was time to reflect again.