With the question posed like this, I have a feeling most folks would select the latter. It’s how you get paid. Today, I want to make the case for having knowledge.
Swimming Against the Current
This semester, I’ve decided to be more explicit with my students about my beliefs on generative AI in the classroom. The short explanation is that I don’t believe it belongs anywhere in education, but I am powerless to stop students from using it. In fact, given the mass adoption of a technology that’s merely 4 years old, I think it’s actually a selling point to host a class that’s free from AI. I even say, “let this be the one space where you don’t use it.”
Unfortunately, despite the average person hating generative AI, I find myself swimming against the current in the classroom. For whatever reason, computer science students haven’t gotten the memo, as many of them even boast about “writing” dozens of apps using the technology.
To make matters worse, pressure is coming from all levels of the university to incorporate AI literacy/fluency in the curriculum in some way. Even if the pro-AI side of the debate is right (i.e., that students should learn to use AI), I find it comically absurd how quickly this adjustment is being made. After all, the vast majority of professors haven’t even begun to adopt a single evidence-based teaching practice from the last 50 years. They’re all still reading directly off slides—ironically, once again adopting technology into the classroom without revising their teaching practices.
Of course, as much as I believe my peers should know better, I can’t really fault students (or graders) for going all in on AI. Their future seemingly depends on it. After all, for every student that falls for some tech industry talking point, there’s another student who is amazed by what Claude Code is able to produce. There is just no way to meaningfully convince them otherwise.
Ultimately, it really does feel like a lost cause to be anti-AI in computer science education right now. Sure, I know several peers who also can’t stand the current state of the field, but we’re considered Luddites in the least charitable interpretation of that term. We’re constantly stuck parrying silly arguments around the future of the discipline.
In Defense of Memorization
One such argument I found myself facing recently came just a day after I stated my position on generative AI. One of my students thought he might be the first person to change my mind, I suppose. After a little back and forth, he asked me something to the effect of: “don’t you think it’s more important to know how to access knowledge than it is to have the knowledge?”
I was somewhat baffled by this question because it’s actually an argument I would have made (and maybe even have) at some point in my life. After all, I have always hated courses that focus on rote memorization. Why would I ever need to memorize a formula when I could easily look it up? In the age of the internet and easy access to information, it seems silly to memorize things.
Yet, today I value memorization. Or as I like to call it: knowing things.
First, if you offload all of your knowledge to a machine, what will you do without the machine? This seems silly because it’s not like the internet is going away, nor is the internet that different from, say, the ability to write something down. However, I feel like there is something to be said about how many of our skills and abilities depend on machines. Surely, an artist can draw without a tablet. How many developers do you think can code without the internet? How about autocomplete? That’s some serious offloading.
And speaking of offloading knowledge to machines, how concerned are you about the “monopoly on knowledge” problem that LLMs present? I am certainly not excited at the prospect of tech oligarchs hoarding all of our knowledge and forcing us to rent it indefinitely. If we’re going to offload knowledge, we should at least be leaving it with libraries, communities, and academic institutions.
Second, if you don’t memorize anything, how can you possibly build knowledge and understanding? For instance, let’s suppose you look up a formula, but you don’t remember what the symbols mean. You’re simply stuck looking them up and piecing together the meaning from scratch. It’s a mess that’s easily solved by having prerequisite knowledge. In fact, this is actually the position I took when debating the student, which went something like this:
If I handed you a book from the 16th century, is that valuable to you? I just gave you access to knowledge, right? Of course not. You would struggle to even read it.
Looking back, I didn’t even need to discuss “ancient” documents. I could have just handed him an academic paper or a book in another language. Having access to knowledge does not automatically grant understanding.
Third, the process of memorizing something actually leads to deeper understanding. In the Japanese course I’m taking, we memorize scripts that we’re meant to reproduce in class. While I’m trying to memorize them, I find that the process goes a lot quicker when I understand the underlying meaning. It’s one thing to just drill words and phrases. It’s another entirely to try to understand what you’re saying. Over time, I often find that something jumps out at me during the memorization process, and I grow a deeper appreciation for what I’m learning. Perhaps the boredom of the process itself induces learning.
I’m sure I could go on, but it’s clear to me that memorization is an obvious first step in learning. I suppose it wouldn’t be the first domain in Bloom’s Taxonomy otherwise.
The Argument for Knowing Things in the Age of Slop
Of course, I understand that the argument for “knowing things” in the age of the internet is a tough sell—one that I’m not sure I could have pitched to my younger self. In the age of generative AI (read: slop), I actually think the case is much easier to make. After all, consider all the ways that “knowing things” helps you combat the onslaught of mis/disinformation produced by generative AI models (e.g., deepfakes, scams, hallucinations, etc.).
In fact, there are dozens of examples of how “knowing things” trumps “knowing how to access knowledge” in the age of slop. For starters, arguing that “knowing how to access knowledge” is more important than “knowing things” requires you to argue that “knowing how to access knowledge” is a skill. Previously, knowing how to use a search engine was a kind of skill, but it was still comically easy. If AI makes accessing information even easier, what skill is there to develop?
And because “knowing how to access knowledge” requires less friction, there is just less learning happening overall. Just consider what academics are doing right now: so many of them have started using AI notebooks like NotebookLM. While I haven’t personally used it, I’m under the impression that it (or at least tools like it) can aggregate literature for you.
To me, this is problematic for a whole host of reasons, but I genuinely think it’s important for you to not let an AI find keywords for you. I spent so much time during my PhD just exploring bodies of literature because I knew what kind of work I wanted to do, but I didn’t know the right keywords. This kind of friction in the learning process is a good thing! I was forced to slog through mountains of literature to find that the thing I wanted to do had a name: value congruence. If an AI had surfaced that for me, what would I have learned? And, could I really call myself an expert?
Also, what feels better: “knowing things” or “knowing how to access knowledge”? Obviously, the process of learning feels worse than just looking something up. After all, friction doesn’t feel good. With that said, “knowing things” certainly feels better. Once again, consider the example of me learning Japanese. Yes, it sucks. I feel dumb all the time, but I’m doing it for the payoff of being able to elevate my experience on my study abroad trips. I want to connect with real people in their native language. To me, that sparks joy. I don’t want to reach for Google Translate every time I want to talk to someone in Japan. Where is the joy in that?
Honestly, even if you’re still not convinced, just consider how preposterous the argument for “knowing how to access knowledge” seems when you account for the fact that knowledge isn’t just some set of facts or objective truths. Most of what you might care to know on a day-to-day basis aren’t facts at all. They’re opinions which you have to evaluate for merit. You can’t just ask “chat” to tell you how to think.
Like, I’m imagining someone opening their favorite LLM and asking it “what is X book/movie/show about?” as if there is some objective lens through which to present the plot. At best, you’ll get a synopsis. Meanwhile, if you ask a person, you’ll get an interesting perspective that accounts for their lived experience. For instance, I remember someone describing Chainsaw Man as something along the lines of “imagine what would happen if an author had psychosis and also hated America.”
The point I’m trying to make is that “knowing how to access knowledge” is not really enough on its own. Even before AI or the internet, knowing how to navigate the Dewey Decimal System is really only part of the problem. You have to know how to read, synthesize knowledge, evaluate arguments, and more. “Knowing things” is, unfortunately, a prerequisite for “knowing how to access knowledge.” So no, I would not say it’s more important to know how to access knowledge than it is to have knowledge, especially in the age of slop.
When Would Access Matter More?
What I find most interesting around this question of “knowledge vs. access” isn’t the question itself but rather a question it raises: “when would it ever be more valuable to access knowledge?” After all, doesn’t the saying go: give a man a fish and you feed him for a day; teach a man to fish and you feed him for a lifetime?
In other words, I would argue that “knowing things” (i.e., knowing how to fish) is always going to be more valuable long term than knowing how to access knowledge (i.e., knowing someone who knows how to fish). Of course, even this analogy sort of falls apart because I’m in favor of communal knowledge (i.e., knowing someone who knows something you don’t know), just not the dystopian future presented by guys like Sam Altman who want to withhold knowledge from us as a “utility.”
Regardless, the only reason we have this relationship reversed in our minds in 2026 is because our entire society is built on productivity. Your output is significantly more important to your boss than your process. And as long as society functions in this way, the student is right: it is more important to know how to access knowledge than it is to have it.
But, that doesn’t mean I have to agree with how society works. And as long as I am an educator, I’m going to focus my energy on making sure students actually “know things.” Sue me.
Hope that was interesting! I feel like this student’s question really bothered me because it reminds me of how dehumanizing society continues to get. Why learn or better yourself if the system demands you to be a cog in the productivity machine, right? Like, we don’t even accept that logic in sports: the process is always more important than the outcome.
Anyway, I desperately want to explore these ideas deeper, but I keep finding myself writing these with a baby on my chest and a jealous toddler lurking. I know I don’t have to apologize for that, but I like to remind folks that I, too, am human. Perhaps I should be spending more time with them anyway!
As always, if you liked this, there is definitely more you can read below:
- Generative AI in Education is Pay-to-Lose
- Why I Rebel Against the Use of Generative AI
- Generative AI Makes It Feel Bad to Be an Educator
Likewise, I’d love if you ran over to my list of ways to grow the site. Over there, you’ll find fun links to things like my Patreon and Discord. Otherwise, take care!
Recent Teach Posts
As we enter the age of slop, I have a question on my mind: how do we assess understanding? I don't answer that today, but I do show you some slop!
Recommended Reading: "To teach in the time of ChatGPT is to know pain" by Scott K. Johnson
Looks like I found another cool article I wanted share. This time it's about the impact of generative AI on education.