|
i want all this poo poo banned
|
# ? Mar 21, 2023 21:17 |
|
|
# ? Apr 29, 2024 02:33 |
|
me too
|
# ? Mar 21, 2023 21:40 |
|
not because i think its going to take over the world or anything silly like that, but simply because other people like it and i am an inveterate, incorrigible Hater. gently caress yoU!!!!!!
|
# ? Mar 21, 2023 21:41 |
|
there is surely no need to have specific reasons to hate on a bit of technology in yospos, especially when it is all the hype. being a touch swept up in it by way of work i do appreciate the balance at any rate.
|
# ? Mar 21, 2023 21:55 |
|
your artificial intelligence is a piece of a poo poo
|
# ? Mar 21, 2023 21:56 |
|
|
# ? Mar 22, 2023 02:48 |
|
Humans not doing much better, downtown Manhattan,
|
# ? Mar 22, 2023 03:42 |
|
https://twitter.com/ChrisJBakke/status/1638323074205614082 lol'd at this
|
# ? Mar 22, 2023 03:42 |
|
MrMoo posted:Humans not doing much better, downtown Manhattan, theres a ton of weird sculptures around here. in the building next door theres a giant bronze ruth bader ginsburg that looks like its melting. down the street theres a giant steel fist. etc
|
# ? Mar 22, 2023 03:44 |
|
welp; ADOBE has gone all-in here it seems - cliff notes their model is trained on free or licensed imagery only so no 'art theft' to get you sued when your million dollar ad campaign launches, a simple interface and direct integration into your usual Photoshop/Illustrator workflow. https://www.youtube.com/watch?v=5cVqYlOpjZs&t=114s I do feel bad for commercial artists here, before you had to buy in to MidJourney or be slightly nerdy and install / tinker with the obtuse Stable Diffusion interface, once publishers get into this it will be a gently caress to get a commission. Of course if you have an exact end result in mind your human artist is still going to be able to feedback and zero in on your ideal result, but this for 95% of book covers / article illustrations? Big oof. edit: this guy is a bit of a fart-huffer where AI stuff is concerned, but its the best breakdown of the stuff in Firefly I could find. NoneMoreNegative fucked around with this message at 11:23 on Mar 22, 2023 |
# ? Mar 22, 2023 11:20 |
|
still quite possible that this turns out fine for artists. like, the creation of photoshop itself made making graphics vastly more accessible to everyone, but it was overall probably good for artists (in fact mostly let more people become very legitimate artists). not like an amateur slamming the buttons will even *recognize* if what comes out works for its purpose the way a professional would, and the amount of human input the things take, and the expectations of the results, will shift accordingly.
Cybernetic Vermin fucked around with this message at 11:46 on Mar 22, 2023 |
# ? Mar 22, 2023 11:44 |
|
or, in general, the quality of work will drop just a bit more as gpt output becomes "good enough", because no one is going to pay to do something well when it can be done poorly for free and there's functionally no difference between the two in terms of monetary returns
|
# ? Mar 22, 2023 13:19 |
|
language models specifically seems more like that, in that the market for word salad has no obvious standards for quality to start with (which also makes it no great loss, the people doing actually good work will not be the ones harmed by that). for visual arts and illustration though i don't think it will even go that far, doing something "good enough" will quickly be seen much like having your kid put photoshops on your website, and as the tools get more sophisticated they will have more and more controls to mess with, in a way that will require just as much expertise.
|
# ? Mar 22, 2023 13:29 |
|
i don't really agree with that. the trend in tool design is simplification. any sort of advanced interface is usually not exposed to the user and you therefore have to be a computer toucher to utilize it (if there's an api at all, big if) most creatives are not very technically inclined, even today. when they are, they might be savvy at actually using computers, but broach any talk about programming stuff and they will get very very mad
|
# ? Mar 22, 2023 13:44 |
|
i work with a number of artists and artist adjacent people (collectors, gallery owners) specifically because the one thing they have no interest in knowing is what buttons to push. it's not a thing that fits into their mental model for the world, and for quite a while that was fine.
|
# ? Mar 22, 2023 13:51 |
|
to some extent one of those where we'll have to wait and see, i don't think we have a very clear perspective on the shape these things will take yet, and it is just very easy to go straight to "this is the end of art" like when cameras were introduced. i have certainly not yet met the artist that truly resents their digital tools, any one individual will not necessarily love struggling with some new dumb thing, but as a group i kind of doubt they'll even be weakened by this.
|
# ? Mar 22, 2023 13:56 |
|
i'm not saying it's the end of art, any more than it's the end of writing. people will still draw and paint and write and create because that's what they want to do, they just will have a much harder time monetizing it, and i think it's going to be almost impossible to make available for free without it immediately getting ripped off by the plagiarism engines on the commercial side, you'll see a general dip in quality across the board, mirroring what you already see out of lovely content mills, because the output is good enough for free that there's no sense paying for anything better
|
# ? Mar 22, 2023 14:00 |
|
yeah. i don't think it's the end of art (or photography), but it is probably the end of getting paid well for it without having to clear a bunch of bullshit hoops first, like getting a master's degree to draw pictures. we were already almost there anyway
|
# ? Mar 22, 2023 14:04 |
|
yeah, well, we're getting into fairly finicky economics there, and being basically optimistic i of course am hoping some of that can change entirely separate from this, so i think we'll have to agree to disagree rather than let this sprawl too much (especially as i will likely look a bit deranged expressing hope that people will get paid).
|
# ? Mar 22, 2023 14:07 |
|
lol. remember songsmith? is there a music gpt model yet?
|
# ? Mar 22, 2023 14:12 |
|
https://twitter.com/duudeede/status/1638461216610824192 Yeah RIP
|
# ? Mar 22, 2023 14:17 |
|
drat that’s too realistic, the cops are literally clones of each other.
|
# ? Mar 22, 2023 14:32 |
|
i did a couple ai trumps in the cspam tuptup thread last night to break in the new bing generator (because it told me it can not make a trumo) they are mostly pretty bad but they made me chuckle idk.
|
# ? Mar 22, 2023 14:32 |
|
the model is helped out a lot by how unnatural his hairdo appears in reality.
|
# ? Mar 22, 2023 14:39 |
|
wake me up when chatgpt can write and train a chatgpt-like ai but better
|
# ? Mar 22, 2023 14:39 |
|
one of the nice things about the gpt models is that they are architecturally limited in what can be represented. there's some simple language games humans can (with some effort) play which a transformer architecture cannot represent (much less learn; there's an asterisk here, but it is off there with the p=np's of the world), regardless of parameter adjustments and total parameter count. which does mean that one can safely disregard anyone pretending that general ai lies down this path, and can also hope that we'll get a much better understanding of what is going on in the models over time, as limits tend to create structure.
|
# ? Mar 22, 2023 14:46 |
|
Cybernetic Vermin posted:the model is helped out a lot by how unnatural his hairdo appears in reality. it actually is not that bad at capturing his likeness, at least in the hair. sometimes it was wildly off, i.e. but generally it was the face that it couldn't really do. some of them are fine but most are not: which are mistakes i would not expect a human artist to make, even intentionally. they'd have to be truly pretty bad
|
# ? Mar 22, 2023 14:55 |
|
since these thing regurgitate poo poo from stuff they learned from the internet, are they going to become shittier and shittier the more they are used because more and more lovely ai generated stuff circle back into their learning corpus?
|
# ? Mar 22, 2023 17:02 |
|
Zlodo posted:since these thing regurgitate poo poo from stuff they learned from the internet, are they going to become shittier and shittier the more they are used because more and more lovely ai generated stuff circle back into their learning corpus? there's a watermark on most diffusion images identifying them as generated (just that one bit, generated: yes), so there's some planning there. for text not as easy, but tbh i think the models can pretty reliably identify useless text by it being *too* surface-level probable, cutting that out of training. all up in the air as time goes on though, if indeed people mix and match using the stuff.
|
# ? Mar 22, 2023 17:08 |
|
i think a lot of that goes out the window once people modify text to be more convincing (again, a big if, because the goal is laziness), but the same applies to images: someone that might not want to reveal that their crap is ai generated will simply crop that out. if they use something like photoshop to clone stamp or whatever it out, then welp, it's useless that doesn't preclude something like steganographic watermarking, of course, but i'm not aware of that occurring anyway (as may be the point, but still. recompression might obliterate that also)
|
# ? Mar 22, 2023 17:15 |
|
Beeftweeter posted:i think a lot of that goes out the window once people modify text to be more convincing (again, a big if, because the goal is laziness), but the same applies to images: someone that might not want to reveal that their crap is ai generated will simply crop that out this is a steganographic watermark, i am not that sure everyone does it, but stable diffusion ships with it and at least some other models have followed suit. perfectly possible to destroy, but that's at least a first bulk filter. it'll be an interesting thing in itself, because to some extent success in filtering beyond that will depend on having the model that made the thing, and as it is in all competitors interest there might be a bit of a pact to create a common service which just tells you how likely models judge it is that it generated the content. the case where all models just pollute each other is more interesting though, the year is 2145, bing still insists that avatar 2 is not out yet.
|
# ? Mar 22, 2023 17:21 |
|
Cybernetic Vermin posted:this is a steganographic watermark, i am not that sure everyone does it, but stable diffusion ships with it and at least some other models have followed suit. perfectly possible to destroy, but that's at least a first bulk filter. it'll be an interesting thing in itself, because to some extent success in filtering beyond that will depend on having the model that made the thing, and as it is in all competitors interest there might be a bit of a pact to create a common service which just tells you how likely models judge it is that it generated the content. There's this Twitter thread about someone asking bard when google would shut it down, it replied that it was already shut down, and then in a further question about this it linked that previous tweet as evidence of it being shut down. That's why I'm asking about this, it seems pretty easy for these things to start huffing their own farts.
|
# ? Mar 22, 2023 17:28 |
|
Cybernetic Vermin posted:this is a steganographic watermark, i am not that sure everyone does it, but stable diffusion ships with it and at least some other models have followed suit. perfectly possible to destroy, but that's at least a first bulk filter. it'll be an interesting thing in itself, because to some extent success in filtering beyond that will depend on having the model that made the thing, and as it is in all competitors interest there might be a bit of a pact to create a common service which just tells you how likely models judge it is that it generated the content. yeah, interesting. i can see that possibly failing if people do quick edits to ai generated stuff to correct obvious mistakes (correct number of fingers, warped faces, poo poo like that). especially if the steganography isn't obvious in that case, it would probably be easy to destroy i'm not sure how other models might respond to that honestly. it seems most of them have the same problems, with stuff like text generation based on just piecing together other images, etc. but in that specific example i don't see how they'd be able to identify that in input if that can't be reliably identified upon output either — maybe OCR, but that's no guarantee of success either. it's not really an image processing model's job. idk, i suppose with time and more complexity these things will be addressed, but it's not like i'm rooting for them per se. right now my entire family relies on creative work
|
# ? Mar 22, 2023 17:29 |
|
Zlodo posted:There's this Twitter thread about someone asking bard when google would shut it down, it replied that it was already shut down, and then in a further question about this it linked that previous tweet as evidence of it being shut down. i saw that too. i think it also misidentified a yahoo news article as a source? i suppose a problem there is that ai text models are going to use things ostensibly authored by humans as a primary source, news organizations specifically. unfortunately they are essentially already content mills and we've seen more than one example of "journalists" just churning out ai crap as their own work (i.e. CNET) because it is easy to game for SEO
|
# ? Mar 22, 2023 17:33 |
|
Zlodo posted:There's this Twitter thread about someone asking bard when google would shut it down, it replied that it was already shut down, and then in a further question about this it linked that previous tweet as evidence of it being shut down. the web searching aspect is already going insane yeah, bing also claims that bard is getting shut down now. but that's in my mind a different matter, these are really interesting demos, but as a product this free-running conversation they stuff web searches into is insane nonsense. i think the tech has a ton of potential for good, and even the crazy tech demos are mostly healthy (let people see and understand where things are at, faced with it with all corners already shaved off there'd be way more dangerous confusion), but there's no way free-running text generation with random prompting shoved in is where it'll do anything useful. also valid: if one was in a position to, just ban it all. but for my sanity i accept some things are going to happen no matter what i do.
|
# ? Mar 22, 2023 17:37 |
|
https://twitter.com/juanbuis/status/1638289186351456257 for reference
|
# ? Mar 22, 2023 17:45 |
|
as a purely probabilistic model needing to answer with no facts provided that is pretty good really
|
# ? Mar 22, 2023 17:47 |
|
Cybernetic Vermin posted:as a purely probabilistic model needing to answer with no facts provided that is pretty good really i mean, the problem is that it is ostensibly working with facts. i haven't tried bard myself (i just joined the waitlist), but it seems to function similarly to bingGPT (i.e. it does a search and works off of the provided results) which, sure, it's impressive in that sense. but you get the same problem: a lot of results there are not going to be valid, and a lot of them will in fact be hallucinations dreamt up by other probabilistic models (or worse, to be honest). it's probably asking too much for it to provide factually accurate results at this point but that nuance is lost on a lot of people. i've also seen examples of people using "analysis" by loving chatGPT as being factual and then reported as such, with absolutely zero shame in it also being reported as a primary source that ties into human psychology a bit too, people are trained in school to treat something facially valid and written in an authoritative tone as being truth without second-guessing anything. they simply accept what they see, especially if it comes from a "trusted" source (i.e. news organizations that have been gutted by cost-cutting and such are increasingly turning to generated content without any editing). it's not good e: grammar Beeftweeter fucked around with this message at 17:57 on Mar 22, 2023 |
# ? Mar 22, 2023 17:55 |
|
yeah, that post was just a joke: if you don't know what x is, but you know it is a google product, what would be your guess when asked when it shut down?
|
# ? Mar 22, 2023 17:57 |
|
|
# ? Apr 29, 2024 02:33 |
|
Cybernetic Vermin posted:yeah, that post was just a joke: if you don't know what x is, but you know it is a google product, what would be your guess when asked when it shut down? in that sense it reporting the same date it generated the result was
|
# ? Mar 22, 2023 17:59 |