> while the chains of thought chatbots produce look like deliberation, research has demonstrated that the bots often concoct them after the fact, reaching an answer by one route but reporting another.
Doesn’t research show humans often do this too? There’s a pretty famous paper from the 70s about that [1], and lots of subsequent evidence. We also have choice blindness [2], we confabulate reasons [3], and we even change our choices (sometimes negatively) after trying to introspect [4].
The BASIC print+goto loop might have been my first ever program too (in the 80s) but one thing for sure for me: it was definitely not “Hello, World”. I and others filled the screen with any number of silly things, random words, jokes and juvenile phrases, whatever, but for some reason, “Hello, World!” was a C thing, and I remember seeing it for the first time in a C class in high school, and then much more often in college.
I have memories of ‘poke’ as well. On the Atari 800 you could poke memory to change the keystroke repeat rate and delay before repeat started, so obviously we would turn down the delay to zero and turn up the repeat to max (it was maybe 60 characters per second), so anyone trying to type couldn’t hit the keys quickly enough to avoid repeating. You’d have to power-cycle the machine to fix it, so it was hilarious to 11 year old me.
Similar to his description of ‘poke’, but different mechanism - using debug.com on an IBM PC jr to edit the Ultimate III save files, and give myself maximum money and lives.
The reason that "Hello, World" is a C thing is that C is the source of the tradition around that phrase. In Kernighan and Richie's "The C Programming Language" printing "Hello, World" is the very first exercise.
Interestingly enough, in the linked text, B is referred as a "language", follow by "the generated code is quite good", presumably referring to the resulting Assembly code.
I disagree as that doesn’t introduce using the B environment with a Hello, world program which is the whole point. After K&R it was the canonical first program.
I think my second BASIC program was the "guess a number" one -- that seemed really popular in intro basic programming things in the 80s also since it showed keyboard input and conditionals and was, uh, fun-adjacent.
and struggled to type this without errors into computers in department stores. I must have read this code somewhere and needed to see it working for myself.
I remember discovering ? as a shortcut for PRINT on the C64. That was when software development as a profession really started looking like a good choice to young me.
In high school, my friend and I were demonstrating to Fred the magic powers of our IMSAI 8080 with ye olde:
10 PRINT "FRED IS A GEEK"
20 GOTO 10
To wit my friend and I would giggle and taunt.
Undaunted, Fred took a look around, found the power button, and turned the machine off. Yes, that power button next to the "DO NOT TURN OFF" sign.
As the monitor and LEDs faded to black, the giggling stopped and my friend and I stood aghast at Fred's treachery.
Fred left, and we were tasked with restarting the machine, getting to the monitor, keying in the hex code of the cassette loader, and reloading BASIC from cassette.
On my Spectrum I came up with:
10 GOTO 20
20 GOTO 10
It would start the Spectrum's "loading screen" border which would then scroll infinitely.
At some point I also opened it up to see what it looked like on the inside, but I managed to damage the keyboard in the process. Thereafter whenever I wanted to play a game (via joystick) I would have to use a piece of wire to manually select the rows/columns of the matching keys to enter the load command.
In C, the classic hello world example gets to demonstrate compilation and defining functions. In BASIC, `10 print "Hello, world"` does neither, so it's not a very interesting storytelling step. Hence, the introductory materials for the two end up chunked differently.
Cute idea, but is this fixed width words, and not fixed width tokens? Isn’t an ‘ing’ ending normal a separate token from the root? Isn’t ‘strawberry’ two tokens?
I think it is: look at the "in" which is likely one token, "strawberry" looks like it is 2x the length of "in". Same for the last word, "ing" seems less compressed than "bear" right before it.
Looking at the first several examples, shorter (one, two, three-) letter words are not stretched to the full ‘fixed’ width. It’s the whitespace around them that is fixed. The only example in the first three screenshots of splitting a word into tokes is ‘vibes’ which turns into ‘vib’ and ‘es’. I’m skeptical of the -ing and -tion words being single tokens as shown.
Why is that interesting? How often does that happen? Has it happened even once? Seems like fantasy; this has been possible for a long time, and despite it being possible, it almost never actually happens. Before AI, kids willing to orchestrate their own education could use the internet to become experts in anything. Before the internet kids could use libraries and books to become experts in anything. But how many take the time and effort to match a 4 year degree? Statistically, approximately zero.
One major problem: how do you get a job for which a degree is required, if you teach yourself via books/internet/AI? How do you get into graduate schools for which a degree is required? How do you envision advanced degrees working without funding or labs or mentors?
All of education is undergoing dramatic seismic shifts because of AI right now. Attempting to insult teachers for talking about it doesn’t reflect on teachers, it reflects on you.
It's interesting because it clarifies that the problem here isn't the kid's education; the 'problem' is that teachers and colleges are upset that they are no longer offering something valuable.
The kids are fine; the rent seekers aren't.
And to your larger point: the entire university/graduate/lab/scientific system was created to formalize what was originally citizen scientists trying to understand the world from their basements. Scientific journals are just a formalization of nerds writing letters to each other about what they figured out. Consider from first principles how much of what exists now actually still serves that original purpose.
You want it to be a jobs program? Sounds good. Take their computers away and make them use pencils and you can create even more jobs.
You want to do science? Figure out how to do science. You want the kids to learn? Figure out how to teach them.
Just don't protect a jobs program under the guise of something nobler.
I disagree that’s the problem, and I don’t think your claim is true. Degrees are still extremely valuable, and income stats alone prove it conclusively. I don’t find your suggestion very interesting, not because colleges are gatekeeping, but for the reason already stated: because most people do not and cannot teach themselves and curate their own educations at a university level, and because even if they did, which they don’t, it still wouldn’t help them get jobs or advanced degrees in today’s workforce. It might be an interesting suggestion if that happened with any statistical significance, but it doesn’t, and isn’t a realistic path for education for society.
Figuring out how to teach kids is exactly what this article is doing. How do you propose to teach kids, when they refuse to put in the work, and reach for AI for everything? Your suggestion to make them use pencils seems naïve/ignorant/obtuse; this article is written by a computer science teacher. You can’t force people to use only pencils for homework (as the article points out), and it would be ridiculous to use pencils for writing computer programs in 2026. That suggestion is a non-starter. FWIW, the impression I’m getting is not that you have any helpful suggestions for education, but that you don’t have experience teaching and don’t understand what’s happening in schools right now.
Who is protecting jobs programs under the guise of something nobler? This seems like a straw man unrelated to your previous comment. I’m getting the feeling you’re anti-education for some reason, maybe have some bone to pick with the university system, is that accurate? Asking kids to teach themselves or forcing them to use pencils are not going to be successful ways to improve education.
You're making ad hominem comments about me; your argument about the degrees is making some heavy assumptions that such things won't be massively changed by AI.
And you're taking the pencil comment weirdly literally, like it apparently went completely over your head. We're clearly talking past each other. I don't see any purpose to talking about this further.
Things are being massively changed by AI. That’s what this article is about, the article you seem intent on poo-pooing. I only asked you a question about your stance, it was not an attack. My question was aimed at trying to uncover why you’ve cast insults toward teachers multiple times, and implicitly the one who authored this article. The only ad-hominems were yours, and they started with your first comment.
What heavy assumptions are you referring to? If you mean assuming that kids probably won’t skip university and instead will be self-motivated enough to use AI non-stop for several years in order to teach themselves enough to get a job as a biologist or a ballet dancer or a civil engineer, and if you mean assuming that companies and organizations probably won’t suddenly start accepting self-AI-taught candidates as being the same as having a degree… if these are the assumptions you’re referring to, then you’re right - I am assuming those things and I believe them. No matter how AI changes things, it’s a pretty heavy assumption you’ve made to think kids will self-motivate an education just because information is available, especially given that was always possible and still chosen so rarely that it effectively doesn’t exist. And it’s a pretty heavy assumption to think that universities and organizations will change together in a way that can vette the merit of prospective candidates any time in foreseeable future. AI will change a lot of things, but you’re underestimating our social structures and what things like the accreditation system do for the workforce and the economy. You can argue it’s got problems, but assuming like you did that AI can replace it anytime soon is a very heavy assumption.
If the pencil comment and suggestion to take away computers wasn’t literal, then you’re right, I have no idea what you meant. Feel free to explain. If you can’t, then I’d agree, there’s no further purpose.
The article reads to me as largely LLM generated. I checked several sections with Pangram and it appears to agree that it’s either entirely AI generated, or at least 100% AI assisted.
I’m fine with the idea of AI assistance, but I agree the quality is low and this needed a bit more editing. I have my own list of gripes that include not defining terms that are used, text for tables and figures being repeated in the body and the captions, contradictions and non-sequiturs, and just AI’s real watermark of getting lost in details, using clever terminology, having a hard time being concise, and being unable to use plain and clear language.
I think they goofed on the conclusion: “These results do not argue against watermarking for provenance. They show that provenance and behavioral stability are separate properties.”
The article’s entire point is that stability depends on provenance techniques, therefore the results purport to show they are not separate properties. Oops.
Also, it’s not AI’s fault that URLs in the references section aren’t clickable. Come on.
Oh it very much depends, your opinion there is certainly not always true and not shared by everyone; you only know where you stand right now with any certainty.
It depends on the teacher. I’ve had good and bad math classes, and LLMs today are quite a bit better than the bad ones. The worst human teacher in my memory didn’t offer interactions. He walked in, turned his back to the class, wrote equations on the board for 45 minutes, and then left. The best math teachers, the ones better than LLMs, are the ones who share the joy and sense of discovery and history of math, and not just the mechanics. But there aren’t that many teachers of that sort.
Learning by using the internet without LLMs is rarely very good, but more often than not in my experience sucks much worse than using LLMs. If you include using public LLMs in internet usage, then it’s not very different from just using LLMs that search the internet. I have heard that a huge swath of today’s high school and college kids are reaching for chatGPT before Google (which is incidentally OpenAI’s goal), and that many of them would rather talk to chatGPT than talk to a teacher. I’m going to refrain from making any claims, but I believe there are a lot of people who disagree with your ranking of the options.
Foreign language learning is one case where I love using LLMs, because it’s not typically an option otherwise. You can practice non-stop and have conversations with someone fluent in a language who will be infinitely patient with your mistakes. This is true of math and other subjects too; using LLMs to practice, so that the human teacher isn’t the bottleneck, to supplement and reinforce the human interactions, is usually better than using the internet without LLMs. The other reason many people prefer talking to LLMs is the lack of judgement. If you aren’t getting it and ask the teacher one too many basic questions, they treat you differently. Sometimes it’s necessary and helpful, and sometimes it’s harmful and takes a long time to change. LLMs don’t do that, they just explain and explain. That lack of judgement is a big reason many people prefer LLM interaction to human interaction.
Okay, sure. The difference between option 1 and option 2 is completely irrelevant to this discussion and I gave what I thought would be an uncontroversial take on it because I thought it didn't matter. The point is at least one of option 1 or 2 is almost always better than option 3. Option 3 is what I and others are arguing against.
Irrelevant? How so? And why did you present two different options if you thought it was irrelevant? I thought you were trying to make a point about human interaction. From my POV, your options 2 and 3 are much closer together than 1 & 2, so if you’re not suggesting that human interaction is preferable to LLM interaction, then what’s your justification for claiming any option is better than any other? I still completely disagree with your clarified point, and suspect many many other people do too, and are demonstrating that preference by using chatGPT for their classes instead of using either the teacher or Google. It’s a simple fact that interacting with an LLM can be preferable to interacting with some teachers some of the time, and LLMs currently often make interacting with the internet more pleasant and more efficient. Your ‘almost always’ still sounds to me like one person’s opinion that is not shared by all.
We're not talking about humans vs LLMs in general, we're talking about humans copy-pasting LLM output as if they wrote it themselves. It's less "LLMs are bad" and more "if you want to communicate something specific to me, LLMs are a bad substitute for your own writing".
Oh I see I’ve misunderstood. Your analogy is restricting the scenario to no interaction with LLMs. In that case, there are still some reasons LLM writing might be prefereable for students: for non-native speakers, and for math teachers new to match developing a new curriculum (sucks, I guess, but it certainly happens), just to name a couple.
One question to ask is why is your contrived LLM scenario any different than a math teacher using a textbook, or district/state worksheets? Math teachers typically do not write the course text or exercises, and never have. Very few math teachers do their own writing.
I’d guess the concern is performance, not what initializer value is used. And performance is a valid concern that is discussed in the proposal, and a reason there’s an escape hatch. Still, it might cause some confusion.
Doesn’t research show humans often do this too? There’s a pretty famous paper from the 70s about that [1], and lots of subsequent evidence. We also have choice blindness [2], we confabulate reasons [3], and we even change our choices (sometimes negatively) after trying to introspect [4].
[1] https://www.researchgate.net/publication/229060046_Telling_m...
[2] https://pubmed.ncbi.nlm.nih.gov/16210542/
[3] https://pmc.ncbi.nlm.nih.gov/articles/PMC5986841/
[4] https://pubmed.ncbi.nlm.nih.gov/2016668/
reply