@glyph Did you quote post something?
Uncategorized
903
Posts
312
Posters
9.5k
Views
-
@glyph I mean, I don't know what to tell you. Are you still really doubting that these things are useful? I've written so many times about this now, are you dismissing it? I can point you to code that I've written over the last seven days that in terms of complexity and utility, is way beyond what we've been able to push out over Christmas. (eg: https://github.com/mitsuhiko/tankgame which is public)
Like, how can you doubt this? It just boggles my mind.
@mitsuhiko amazing. you plagiarized a game-jam game in about the amount of time that a game jam usually takes to run. truly our society will be revolutionized
-
@mitsuhiko amazing. you plagiarized a game-jam game in about the amount of time that a game jam usually takes to run. truly our society will be revolutionized
@glyph Please tell me what is plagiarized and also no I wouldn't be able to do it without an LLM over Christmas while also working on actual work. I just couldn't have done it. You might be able to. I can't. And that's a pretty big difference.
-
Being close with someone who has gone through arduous kratom detoxes more than once, it became something I would not touch for love nor money.
I don't care if it is "natural" -- I've seen it wreck people's lives. I would never blame an addict, but there are definitely many readily-available pitfalls in the world, and peer pressure is a thing, even with adults.
Having seen starry-eyed, vacant-looking acquaintances rave about LLMs as the new messiah or some such sh*t, I avoid those assiduously as well. Too culty! It won't turn out well.
-
@glyph Please tell me what is plagiarized and also no I wouldn't be able to do it without an LLM over Christmas while also working on actual work. I just couldn't have done it. You might be able to. I can't. And that's a pretty big difference.
@mitsuhiko re: plagiarism: of course I can't tell you exactly what was plagiarized here. it's *extremely* hard to even track provenance of what an LLM's 'inspiration' was, let alone to determine specifically if it made a sufficiently exact replication of training data that direct copyright litigation would be feasible. you don't know who contributed the training data that made this work possible. in my opinion, it's inherently plagiarism.
-
@mitsuhiko re: plagiarism: of course I can't tell you exactly what was plagiarized here. it's *extremely* hard to even track provenance of what an LLM's 'inspiration' was, let alone to determine specifically if it made a sufficiently exact replication of training data that direct copyright litigation would be feasible. you don't know who contributed the training data that made this work possible. in my opinion, it's inherently plagiarism.
@mitsuhiko re: you "couldn't have done it", sure, maybe. that's subjective! which is the exact thing that I said would not move the argument forward. so you're just performing an argument from incredulity here. here's a counterpoint: I asked an LLM to help me with a data structure problem and wasted about a week on useless garbage output. We are now at net zero utility between the two of us, Q.E.D.
-
@mitsuhiko re: you "couldn't have done it", sure, maybe. that's subjective! which is the exact thing that I said would not move the argument forward. so you're just performing an argument from incredulity here. here's a counterpoint: I asked an LLM to help me with a data structure problem and wasted about a week on useless garbage output. We are now at net zero utility between the two of us, Q.E.D.
@mitsuhiko this is the definition of sample bias (which I did my level best to explain in exhaustive detail in https://blog.glyph.im/2025/08/futzing-fraction.html ). It wastes time sometimes, it saves time sometimes, it helps with learning sometimes, it helps with anti-learning misinformation and incorrect conceptual models sometimes. Is it a net benefit or not? I don't think it is, but I can't prove it! It's kinda incumbent upon boosters at this point, given all the info we now have on its risks!
-
@mitsuhiko re: you "couldn't have done it", sure, maybe. that's subjective! which is the exact thing that I said would not move the argument forward. so you're just performing an argument from incredulity here. here's a counterpoint: I asked an LLM to help me with a data structure problem and wasted about a week on useless garbage output. We are now at net zero utility between the two of us, Q.E.D.
@glyph You’re arguing against a strange strawman. On the one hand, you claim this is useless; on the other, when I point out that it’s useful to me and show concrete output that has been genuinely valuable, you dismiss it as something else entirely, and apparently plagiarism.
I get the impression that this is upsetting to you, or that you're simply uncomfortable with people using it. What puzzles me is the complete disregard for the evidence being presented, because at this point it doesn't seem grounded in reality.
I think there's nothing I could tell you that would convince you that this is useful to me.
-
@glyph You’re arguing against a strange strawman. On the one hand, you claim this is useless; on the other, when I point out that it’s useful to me and show concrete output that has been genuinely valuable, you dismiss it as something else entirely, and apparently plagiarism.
I get the impression that this is upsetting to you, or that you're simply uncomfortable with people using it. What puzzles me is the complete disregard for the evidence being presented, because at this point it doesn't seem grounded in reality.
I think there's nothing I could tell you that would convince you that this is useful to me.
@mitsuhiko I find it upsetting because you're strolling right past all the points I'm making, ignoring the parameters I tried to set on the discussion, and embodying the EXACT THING that I pointed out in the top post. I said that LLM users say "sure it has problems, but I can handle them". And you are responding to that by saying that you've pointed out the addictive tendencies but you still use it because you see benefits. That's the thing that I was saying! That's my worry!
-
@mitsuhiko re: plagiarism: of course I can't tell you exactly what was plagiarized here. it's *extremely* hard to even track provenance of what an LLM's 'inspiration' was, let alone to determine specifically if it made a sufficiently exact replication of training data that direct copyright litigation would be feasible. you don't know who contributed the training data that made this work possible. in my opinion, it's inherently plagiarism.
@glyph @mitsuhiko The way to avoid plagarizing is simple: stop using AI, and use your own damn brain.
2026 is the year of the human, didn't you hear?
-
@datarama @xgranade I have other anxiety issues (OCD does not really resonate broadly, but anxiety+ADHD can produce a similar effect) and I sometimes feel waves of INTENSE rage when I open up my Apple TV and it's playing a loud trailer for a show I don't really want to get into, or launch Steam and find that it's forgotten the preference to open to the Library and not Store tab. We need strict regulation for ALL of these attractive "algorithmic" nuisances
-
@mitsuhiko I find it upsetting because you're strolling right past all the points I'm making, ignoring the parameters I tried to set on the discussion, and embodying the EXACT THING that I pointed out in the top post. I said that LLM users say "sure it has problems, but I can handle them". And you are responding to that by saying that you've pointed out the addictive tendencies but you still use it because you see benefits. That's the thing that I was saying! That's my worry!
@mitsuhiko As far as the strawman, let me try to explain again. If I have said it's "useless" (a word I try not to use, but I might slip up here and there—in this discussion, I describe it as *having produced* useless output for me, which is just a literal thing that happened, not a description of the model overall) what I am referring to is the *overall cost/benefit* not necessarily being positive.
-
@glyph Please tell me what is plagiarized and also no I wouldn't be able to do it without an LLM over Christmas while also working on actual work. I just couldn't have done it. You might be able to. I can't. And that's a pretty big difference.
@mitsuhiko @glyph And cocaine can give you a lot of energy. What's your point?
-
@mitsuhiko As far as the strawman, let me try to explain again. If I have said it's "useless" (a word I try not to use, but I might slip up here and there—in this discussion, I describe it as *having produced* useless output for me, which is just a literal thing that happened, not a description of the model overall) what I am referring to is the *overall cost/benefit* not necessarily being positive.
@mitsuhiko I didn't find your specific example particularly impressive but let's ignore that. Pretend that it's great. How are you measuring that *against* the failed starts, the disinformation, the papering over boilerplate, the repetition due to small context windows, all of the *well known and extremely widely discussed* problems that this technology has? How do you know that it is *overall* saving time? If it saves *you, personally* time, how are you measuring that against social harms?
-
@mitsuhiko I find it upsetting because you're strolling right past all the points I'm making, ignoring the parameters I tried to set on the discussion, and embodying the EXACT THING that I pointed out in the top post. I said that LLM users say "sure it has problems, but I can handle them". And you are responding to that by saying that you've pointed out the addictive tendencies but you still use it because you see benefits. That's the thing that I was saying! That's my worry!
@glyph a lot of things that are deeply fun have those properties. Computer games, even going to gym can be that way. It’s a very exciting time and it’s deeply enjoyable to play with these things. While simultaneously being useful. I’m not sure how I can break that to you.
-
Needless to say, if somebody time traveled back to the 1950s and tried to explain this to the people back then, they would have thought you were crazy...
-
@glyph a lot of things that are deeply fun have those properties. Computer games, even going to gym can be that way. It’s a very exciting time and it’s deeply enjoyable to play with these things. While simultaneously being useful. I’m not sure how I can break that to you.
@mitsuhiko now you're just doing this dril tweet https://en.wikiquote.org/wiki/Dril#:~:text=the%20wise%20man%20bowed%20his%20head%20solemnly
justifying your addiction with moral relativism and an appeal to a benefit that I do not think is, on net, good.
I think that as a society we've got our arms around gaming & the gym, there's plenty of data about how addictive those things are. (And also about how "computer games" is a pretty big bucket, where you can find a tremendous amount of gamblification right now, which is just as bad if not worse as LLMs)
-
The vast majority of LLM usage isn't even going to be voluntary or self-rationalized.
Google is by far the world's most popular search engine, used by literally billions of people per day and Google is busy rolling out their excretable "AI overviews" globally.
The vast majority of "LLM users" couldn't tell you what an LLM even is and are not entering any sort of deliberate decision to engage with an LLM or really thinking about it at all.
They're just going to be searching for something and then reading the thing that is given to them at the top of the results. They are going to be funneled to an LLM when they seek out customer service or technical support or apply for a job or accept the 'assistance' jovially offered to them in their word processor or PDF viewer or right in their operating system.
You can can convince someone to not use ChatGPT perhaps, but I'm really not sure what anyone is going to do about the seemingly universal goal of every technical giant on earth to redefine how we interact with information through the interface of AI.
In a not too distant future it will become very difficult for the average person to know where AI even begins and ends.
-
@mitsuhiko now you're just doing this dril tweet https://en.wikiquote.org/wiki/Dril#:~:text=the%20wise%20man%20bowed%20his%20head%20solemnly
justifying your addiction with moral relativism and an appeal to a benefit that I do not think is, on net, good.
I think that as a society we've got our arms around gaming & the gym, there's plenty of data about how addictive those things are. (And also about how "computer games" is a pretty big bucket, where you can find a tremendous amount of gamblification right now, which is just as bad if not worse as LLMs)
@mitsuhiko one of the reasons I find this so upsetting is that your argument here is so obviously missing the point that it makes me scared that these things can so terribly damage your metacognition that this seems like reasonable things to say. *I* still want to experiment with them to develop a better understanding, and this kind of public rhetoric makes me feel like I might be poisoning my own brain to do so!
-
@mitsuhiko one of the reasons I find this so upsetting is that your argument here is so obviously missing the point that it makes me scared that these things can so terribly damage your metacognition that this seems like reasonable things to say. *I* still want to experiment with them to develop a better understanding, and this kind of public rhetoric makes me feel like I might be poisoning my own brain to do so!
@mitsuhiko I could *easily* accept an argument like "we all have to make decisions under uncertainty and *in my experience*, accounting for subjective distortion as best I can, there has been a big net benefit. we're going to have to agree to disagree until someone does a more comprehensive study; I'll gather more data on my own use in the meanwhile"
but your insistence that I recognize these anecdotal examples (which I *already acknowledged repeatedly*) as *proof* of net benefit is scary
-
@glyph Absolutely, but even in that we can make out the general shape of the thing. We know this current model is not economically sustainable by any party. For users, the result will be inability to access models, or paying cripplingly high prices to do so. Option 1 will elucidate their inability to function without the models, and option 2 will impose the kinds of costs that look like rock bottom in other addictions.
@mttaggart @glyph If there is such an unwinding I think users that can't afford premium service providers will fall back to free/subsidized providers and tools that run on-device. A whole spectrum rather than a binary have / have not.
Gli ultimi otto messaggi ricevuti dalla Federazione
-
@evadm qui non sarà (fe)diverso allorquando questi spazi saranno rilevanti. Per ora ci salva la nostra irrilevanza .
-
@afreytes brilliant!
-
@GustavinoBevilacqua @marcoboh @lysander @thatgiga @LaVi @matz @ju
me le devo aggiungere alla lista di scoperte sconvolgenti
-
@GustavinoBevilacqua @marcoboh @lysander @thatgiga @LaVi @matz @ju
(questa del facocèro però mi sa che l'avevo già sentita qui)
-
@GustavinoBevilacqua @marcoboh @lysander @thatgiga @LaVi @matz @ju
what the cazzo, non anche quel povero animale, che v'ha fatto di male
-
-
@lysander @thatgiga @LaVi @Uilebheist @marcoboh @matz @ju
> TUTTI GLI ACCENTI SON BELLI IN REALTÀ
avevo letto inizialmente «tutti gli accenti sono belín» e si però ma
-
@marcoboh@devianze.city @thatgiga@mastodon.social @oblomov@sociale.network @LaVi@livellosegreto.it @Uilebheist@polyglot.city @matz@gts.matteozenatti.net @ju@gts.nugole.it
Vesto vì è bellissimo, lo adoro!
L'accento livornese in generale è un sacco intrigante.
Un po' tutte le varianti toscane lo sono, a dire il vero.
TUTTI GLI ACCENTI SON BELLI IN REALTÀ
