That said, maybe we don't train models on living artists or watermarked images, ok? It just splashes it with turds and vomit, IMHO. Like what happened to Cleverbot. Gunked up with trolls and idiots, and technically non-consensual too. Maybe we stop, eh? So tasteless.
This one example? Fine. Guy's dead, it's a dedication. Whatever.
I mean, don't train models on watermarked images because it makes the model reproduce watermarks which degrades the quality of the model, but intellectual property as we see it has got to go, it's not healthy for humanity. The illegal thing should be closed sourcing a model, not building it off "copyrighted" images.
We need UBI/social safety/cultural reform to support artists in doing art, a thing that's deeply important. We don't need IP laws that allow people to stifle creativity in order to collect rents on things that can be infinitely replicated for free.
> We need UBI/social safety/cultural reform to support artists in doing art, a thing that's deeply important. We don't need IP laws that allow people to stifle creativity in order to collect rents on things that can be infinitely replicated for free
Which do you put in place first?
I think this is a bit of the issue with this argument. As automation does eat at jobs, you can say that the solution is UBI or something like it, but since that's not there, right now and until it happens (which could be never), it will affect people's livelihood, and it's fair for them to be pushing back.
If such social foundation was already here, I don't think people would be complaining as much, they'd probably be joyful that they'd no longer need to work as hard.
If artists cared enough, they need to make an union of some sort and lobby for non synthetic art labelling for protection like big Diary and get AI generated art to be branded like GMO foods.
TBH Kim Jung Gi drawing process sometimes feels closer to watching an AI creation than a person.
I feel bad for anime artists (well all artists), they work particularly hard. But there are a lot of good manga drawn by mediocre artists, if there's option to change them stylistically closer to much more proficient masters (like a video game remaster), I'd do it in a heart beat. There's going to be audience out there for OG Tomohirio One Punch Man squiggles, but there's a reason it gained mainstream popularity after being redrawn by Murata.
I'm also just waiting for AI to generate music based off the few particular songs I like.
You can be negative or enthusiastic about this. Either way the trend indicates the impending death of an entire occupation of creativity. I envision an AI that can replicate Kims art and replicate it better and faster.
Now for the first time in my life I can see the future beyond science fiction. Where machines are capable of doing everything humans do, and do it better.
The outrage is a violent attempt to stop the inevitable. I sympathize. Because artists won't be the only victims of this change when it's all over.
No. But there are people with ambitions to create robots that will take over my job and yours and everyone's. It's looking more and more like the inevitable future and all people talk about in the comments are trivial things like "intellectual property"
No. But there are people with ambitions to create robots that will take over my job and yours and everyone's.
Sure hope so. Then I can move even closer to self-actualization than I am now.
If there had ever been a single inflection point in history where Luddism was the right answer in the long run, I'd be more sympathetic to the pessimistic point of view. As things stand, though, doesn't their argument amount to a special pleading fallacy? "It's different this time!" No... no, it's not.
Nothing is different. Everything is the same. You're just blind to what is happening around you.
>Sure hope so. Then I can move even closer to self-actualization than I am now.
Yeah and then you starve to death because you have some measly salary. Who says you're the owner of the robot who took over your job? It may be possible, but thinking this is the only possibility is not intelligent. Every realistic possibility must be considered.
With wealth inequality as it is right now, a Few plutocrats reaping most of the benefits of AI with the majority of people scraping by is not just the more realistic possibility, it is also remarkably similar to the current reality.
>If there had ever been a single inflection point in history where Luddism was the right answer in the long run, I'd be more sympathetic to the pessimistic point of view. As things stand, though, doesn't their argument amount to a special pleading fallacy? "It's different this time!" No... no, it's not.
Oh there are tons of examples. You just view history through a biased lens. You're unable to see the times when technology did harm the population. I utterly guarantee that your conclusion is not data driven. You simply are unconsciously picking and choosing the data to support your optimism.
A big portion of the wealth inequality we see today is the result of a century of technological advancement. But you're partly blind to it because you're on the positive side of the wealth inequality equation.
As for "wealth inequality," because of capitalism, our homeless people live better than kings and princes did a few hundred years ago. I'm OK with that.
Able? The examples are obvious. I can cite them though. Once I cite them you should should 180 degree your opinion. But of course you won't. You're so biased you can't even see the obvious examples all around you:
https://web.colby.edu/st297-global18/2018/10/30/has-technolo...
Obesity epidemic caused by ultra processed food technology. Industrialization and the negative effects. "At least they're not starving and living better than kings" is what I imagine coming out of your privileged mind.
https://www.europarl.europa.eu/RegData/etudes/IDAN/2018/6190...
Fake news, misinformation created by social media technology. Before free press and technology people were simply less informed. Now, the current generation of people are likely the most informed but also the most misinformed people who've ever lived. People think they live in a more dangerous world, but in reality crime has gone down. This is all because of the current news cycle.
https://www.icanw.org/catastrophic_harm
Nuclear weapons. The discovery of the relationship between energy and mass and other advancements in physics and chemistry led to the development of technology that enables humanity to literally destroy the world if they chose to. This example literally expands the spectrum about the amount of harm technology can do. In less then a day ICBM and nuclear bomb technology can level the world.
https://en.wikipedia.org/wiki/Chernobyl_disaster
Nuclear power melt downs. There are several. This is the most famous one. Obviously there's nuance to this and I will address it. The net benefit of nuclear power is better then fossil fuels in terms of death and destruction when taken from a statistical standpoint. But you still cannot ignore the fact that a single mistake leveled a whole town. Blindly charging forward because you trust the numbers is equally ignorant.
Need I go on? There are OBVIOUS benefits of technology I would be blind to not to see it. But I'm also able to see the OBVIOUS negative side as well. The nuance.
I would be wrong if I said you didn't know about each of these things. You know it. These t...
The "plastic crisis" is not an example where Luddites were right. At the time it seemed like a good idea to make everything out of plastic, and it probably was considering that we'd have had to cut down an assload of additional old-growth trees if plastic packaging hadn't come along when it did. Overpackaging and irresponsible waste management suck for sure. But legislation and innovation, not wrecking shit, are the answers.
Fossil fuels were not an example where Luddites were right. They have lifted billions out of poverty over hundreds of years at a comparatively small cost, to say nothing of saving countless millions from freezing to death in the dark. Now it's time to leave them behind, and that's just what we're doing.
Rain forest destruction is not an example where Luddites were right. I don't even know where you're coming from here. It should be stopped, but the people saying it should be stopped aren't Luddites.
Obesity and misinformation are not examples where Luddites were right. Again, you're just flailing here, pulling random crap out of your ass. Modern farming practice has saved billions (ever heard of a fellow named Borlaug? It's Google time.) And despite the growing presence of extremism caused by bad actors taking advantage of the same technology that's available to the rest of us, the average person on the street is better informed than ever. If they aren't, it's not the media's fault, it's theirs alone.
"Negative effects of the Internet" reported on an Internet quack medical news site. Okey-dokey, then.
Nuclear weapons put a stop to increasingly-destructive world wars by doing the unthinkable: making the leaders fear for their own lives. That's a boon unmatched by any other in post-medieval history. And nuclear technology could have saved us from countless other ills, if it weren't for mindless FUD from celebrities and politicians with no sense of risk assessment.
Chernobyl? See above. Fossil fuel power sources have killed unimaginable magnitudes more people than nuclear power plants ever could, which is one reason why we need to move past them. But they do it in boring, remote places like hospitals and cancer wards, and they do it years after the harm is inflicted. Without scary green blobs of molten radioactive crap to show on TV, nobody including you cares. News flash: faulty reasoning leads to poor outcomes, (no) film at 11.
Whether you should "go on" or not is up to you. You appear to be addressing a question that wasn't asked, or that was asked by someone who isn't here.
You're just talking about hypotheticals. Saying hey this technology is good because the hypothetical is much worse. Well the hypothetical is just a random guess. As i said you're rationalizing.
>The "plastic crisis" is not an example where Luddites were right. At the time it seemed like a good idea to make everything out of plastic, and it probably was...
"Seemed like a good idea"? and "probably right"? Wow. you have no confidence in what you're talking about here. Your language implies that you AREN'T even sure about the hypothetical alternative if we didn't take the plastic route. The rest of the argument is basically just a rant on how to "solve" the problem and in no way a counter to the fact that plastic technology has harmed the planet. You ingest a credit card worth of plastic per week thanks to the "plastic crisis".
>Fossil fuels were not an example where Luddites were right. They have lifted billions out of poverty over hundreds of years at a comparatively small cost, to say nothing of saving countless millions from freezing to death in the dark. Now it's time to leave them behind, and that's just what we're doing.
More hypothetical's. Nobody was lifted out of poverty. What happens is that food production increased so the planet can support a bigger population. Without that production increase those people wouldn't have existed in the first place. Society existed fine at lower pre-industrial state with a much lower and sustainable population just fine without fossil fuels.
This discussion of hypotheticals is a form of rationalization. There is literally no denying that each of the technologies I list here contributes a negative to society.
>Obesity and misinformation are not examples where Luddites were right. Again, you're just flailing here, pulling random crap out of your ass.
No. Nothing's pulled out of my ass. It's just you're unable to comprehend anything. Let me spell out to you what I'm talking about. I'm talking about INDUSTRIAL FOOD PROCESSING. Not industrial farming. Farming doesn't cause obesity. Food processing does when you strip a food product down to levels of unnatural purity. Go read it again. You're such a genius. Unbelievable.
>"Negative effects of the Internet" reported on an Internet quack medical news site. Okey-dokey, then.
Bro. Everybody on the face of the earth knows the negative psychological effects of social media. The negative health effects of people spending too much time gaming on the internet. Everybody knows this. There's a million sites talking about this. It's OBVIOUS. You want a "better" source? Here: https://web.stanford.edu/~gentzkow/research/facebook.pdf random research paper from stanford. But this is besides the point. You're just arguing to stay alive right now.
I simply "cited" a source to show you how it CAN be cited. But literally you pop up here laughing at the source? It's like I proved to you with a dictionary that O-N-E spells ONE and you just laugh at my source and don't even believe that O-N-E spells one.
Wow just wow.
>Nuclear weapons put a stop to increasingly-destructive world wars by doing the unthinkable: making the leaders fear for their own lives. That's a boon unmatched by any other in post-medieval history. And nuclear technology could have saved us from countless other ills, if it weren't for mindless FUD from celebrities and politicians with no sense of risk assessment.
This is next level mind twisting. The amount of bias you have to have in order to prove your point must be so high that you can even justify nuclear weapons as good. You're too far gone man. Leaders fearing for their own lives as if leaders were never ever under any sort of threat to their lives befor...
>That's not really what this site is for. Your talents would be better put to use over on Reddit, or maybe Fark.
No. You don't get it. It's literally impossible to change a mans opinion. Everyone is too biased and they have their pride tied up with their arguments. Dispassionate logic is a fantasy.
Changing someones opinion is a sign of an effective discussion. Which is basically what NEVER happens on HN. HN is just people patting each other on the back while never effectively conveying information or changing someones opinion. If your opinion changed or even if you got defensive around your position it means you learned something. The discussion was effective and Informative.
That is the purpose of HN. It's just too bad being informed that you're utterly wrong feels bad and is seen as a bad thing. It shouldn't be this way. Realizing you're wrong should be a form of enlightenment.
Also it's not a skill. Being informative is simply about being MORE correct and right then the other party and Informing them about it. Which is what happened to you.
If 'robot's could imitate and exceed people's actual life and being --what should we do about people, virtualize all of them, most of them? They're better off living a virtual life than life in dirty reality.
Creating a specific model from one artist shortly after their death is just plain gross and deserves backlash. It's not a good proxy to talking or thinking about these issues more carefully.
I think there is a big difference between feeding an AI model a wide swath of an art style like anime and creating and using a model.
Taking one persons particular work and style and copying in competition or near competition with that person is something that I don't think is inevitable and something I would be happy to see the expansion of legal protection for.
I know I will get downvoted since HN hates legal protection for artists, but I just don't see the societal benefit from allowing an individual human to be replicated by AI without their consent.
I sympathize with you and I won't vote you down. However I do think that the issue is much bigger then a moral issue. Realize that AI art will become better then human art. It's the trend.
If AI art is better than what humans can make then this is more than a simple moral issue.... basically the inevitable future is that the entire occupation of "art" will be steam rolled and 99% of what's out there will become computer generated. Morals be damned, it won't even be a factor. A bunch of scholarly types trying to quantify the rightness and wrongness of something isn't going to mean jack when the power to do create art is so easy and trivial.
In fact the morals will shift with the culture and technology and it will be a violent shift. People getting angry here is just the beginning of bigger conflicts in the future. If AI keeps improving things will stabilize into an age of humanity where original art is but a cheap commodity, but that age is not one that will arrive soon, the step before that is violence.
The even scarier part is that it's not just art that's being targeted here. Art is just one of the first victims.
That's what they said about sheet music. Bring music into the home was not sometime artists who performed live wanted any part of because it was feared that people wouldn't want to go out. The same was said about recorded music, radio.
In the end it allowed more wealth and created more demand for music. There will be more demand for art.
Art demand is already saturated. It's in everything you see, from websites, to graphics for news stories to the industrial design of all the products you see. AI art is not going to cause more demand for art given that art is everywhere.
More demand for art is like saying more demand for air, more demand for oxygen. Demand is saturated and so is supply. What's happening here is similar to outsourcing. Jobs are going to another place, just not a real place and it isn't people replacing your job.
Your analogy is biased. You picked a specific example of new technology being "not bad". How about picking an example of technology being bad? Plenty of analogies in the opposite direction. Singular examples and analogies are not proof of anything.
Demand for art is far from saturated. Most people have vivid imagination and would like some physical representation of that. For the dozen of amateur creatives there are nearly as many who wish they could get access and have the time to learn a way to express.
Anyway I think the scare about AI generated art is misplaced when it comes to art as a general concept. It is after all merely a tool and a tool can suppress art. Plenty of abstract art was already not limited to technical mastery and could have been made by a computer yet abstract art didn’t disappear and artists weren’t displaced.
The Japanese and Korean illustrators scare however is not misplaced. It comes from how the industry is structured. For every credited artists or directors, you have a whole team of assistants and illustrators who have a technical role and produce parts of the image imitating the desired style. This, I could see replace by AI and as it is the entry level position for a career in the field, it would significantly modify its functioning. These people are scared for their livelihood and they probably are right to be.
> Demand for art is far from saturated. Most people have vivid imagination and would like some physical representation of that. For the dozen of amateur creatives there are nearly as many who wish they could get access and have the time to learn a way to express.
That's not demand for art. That's demand for learning to become an artist. It's like saying people wanting to become farmers is equivalent to more demand for food. No. Related? Yes. But they are not the same.
>Anyway I think the scare about AI generated art is misplaced when it comes to art as a general concept. It is after all merely a tool and a tool can suppress art. Plenty of abstract art was already not limited to technical mastery and could have been made by a computer yet abstract art didn’t disappear and artists weren’t displaced.
You can think this all you want. But you can also project the current trends of technology into the future and envision a future where AI creates art better and faster then ANY human on the face of the earth. Then take that AI and apply it to programming, decision making, car driving and every other occupation that right now only humans can do.
Notice how I just made a realistic projection? I didn't say whether it was scary or horrible. I just made a realistic projection and THAT projection itself should give ANYONE pause about the future of humanity. You are way to optimistic. I leave it to you to fill in what the projection means for us.
A lot of people are just like "Oh we'll just switch to UBI, and AI will create a utopia". Yeah, that's not a realistic projection. Unlike my projection, this is just a random shot in the dark. Given that utopia and UBI never successfully existed on such a massive scale yet, such a transition should also give you pause. Additionally UBI is dangerously similar to communism which is in itself questionable.
>The Japanese and Korean illustrators scare however is not misplaced. It comes from how the industry is structured
No it's from how reality is structured. If some machine does your job better than you then you become redundant. No amount of industrial structuring can make this FACT not true.
>These people are scared for their livelihood and they probably are right to be.
You should be scared as well. AI targets almost every job. Art is just the first to fall.
Well, you haven’t. There is absolutely nothing in the current AI state of the art which points towards that. As many as been telling you, you are scared of ghosts.
Did the loop make workers disappear? Did cameras made us stop painting?
What you wrote were written in exactly the same way for all these innovations. You will pass, progress will march on.
>Well, you haven’t. There is absolutely nothing in the current AI state of the art which points towards that. As many as been telling you, you are scared of ghosts.
It's called logic, there are no ghosts here. If improvements arrive incrementally as they have been for the last decade with milestones like Dall-E, GPT-3, Alpha Go, Deep Blue. These events form what's called an "upward trend". If you project that trend following mathematical statistical procedures with a trend line, you can "project" that line into the future. That is called a "linear projection".
That trend is far more mathematical, far more logical then any other evaluation you can come up with. Everything you have are just qualitative opinions. Literally I have time on the X-axis and "amount of things computers are able to do almost as good if not better then humans" on the Y-axis. That trend line flows back from, the inception of computing. Not just a decade of data but half a century. My conclusions are DATA driven. Your conclusions are just gut feelings.
If you deny this you are illogical. There is no other way of putting this.
>Did the loop make workers disappear? Did cameras made us stop painting?
Wtf is the loop? Did automation in factories make workers disappear? Did cameras make us stop painting?
The answer is yes. Many painters who use to go to peoples houses to paint portraits.... most of those occupations HAVE disappeared. Same with factory workers, automation wiped a lot of them out (aka disappeared). You can see the results of it today... the wealth inequality we see in the world is partly the result of technology.
You are in huge denial if you think technology doesn't make occupations disappear. This kind of thing is a Well known economic phenomenon NOT exclusive to AI. When technology improves workers in related fields become less. There are loads and loads of quantitative line graphs with trend lines that show this and economists who can back it up.
>What you wrote were written in exactly the same way for all these innovations. You will pass, progress will march on.
What do you mean I will pass? Of course progress will march on. There's NOTHING anyone can do to stop it. AI will steam roll entire industries including eventually itself. You already have ML building models and ML training itself.
You're like a climate denier. I'm saying global warming is inevitable. It will happen. There is nothing left humanity can do to change the current trend. You're saying there is no trend. No evidence. No data within the last century at all. Not to be insulting but this is total and complete blindness. Literally I am describing to you trend lines and data points and you're unable to see how biased you are.
If some machine does your job better than you then you become redundant.
Yeah, just think of all those farmers who had no choice but to commit suicide when threshers, steam engines, and milking machines were invented.
You're confusing a boon with a tragedy. It's a GOOD thing to be made redundant by automation. Now you can do what you were supposed to be doing all along... whatever that is.
>Yeah, just think of all those farmers who had no choice but to commit suicide when threshers, steam engines, and milking machines were invented.
Yeah. Many farmers did become redundant. You're not thinking clearly.
>You're confusing a boon with a tragedy. It's a GOOD thing to be made redundant by automation.
You're an extremist thinking all technological advancement is a boon never considering the nuances of situations. It can be good and it can be bad. Obesity for example is the result of improved efficiencies in automated manufacturing of food.
> Now you can do what you were supposed to be doing all along... whatever that is.
What's funny is you don't even know what they're suppose to be doing. What's a human being suppose to do when every ounce of his capability is rendered redundant by machines? What happens to the economy how does it work? You don't know and you think it's a good thing. That's bias right there. Not knowing something and thinking it's good AS IF you know it. You can't know something is good without knowing what that something is.
You know how the economy works right? Let me tell you what it actually is. It's an UNKNOWN thing. Never in the history of the human race has this type of economy existed where most humans no longer need to be productive. If you think everyone is going to have equal access to the benefits of AI and automation history says this isn't what typically happens.
> Now for the first time in my life I can see the future beyond science fiction. Where machines are capable of doing everything humans do, and do it better
Maybe one day, but we're still far away. To contextualize, Stable Diffusion needed 5.83 billion images to be trained on, and it also uses another model CLIP that needed 400 million labeled images to be trained.
That means that all tasks where we don't have hundred of millions or more of available accumulated datasets are really far off from being learnable by current ML models.
To contextualize even further. Stable diffusion will fit on any persons computer AND the time it takes to run is minutes.
But that's just about stable diffusion. You need to measure the speed of how fast stable diffusion came to be. 3 years ago all of this stuff was thought to be impossible. Now it's a SAS. What does the next 3 years bring? Contextualize that.
Nobody knows when that day will come, BUT within our lifetimes is a very realistic bet given the speed in which the ML industry is changing and growing.
Trained Stable Diffusion can, but training it costed 600 000$ and took 150 000 GPU hours.
Also, Stable Diffusion isn't a revolution in model architecture, it's still a deep neural network. There's been a lot of innovation in how you can best apply DNNs to problems, and that's where CLIP and stable diffusion innovate, but also a lot of the advancements has been purely due to sheer brute force from funding more compute resources to train deeper networks using larger datasets.
I'm optimistic, but I'm also trying to contextualize to have the right expectations.
> Trained Stable Diffusion can, but training it costed 600 000$ and took 150 000 GPU hours.
So? The iphone as it is today took EVEN more man hours and billions of dollars to create. Yet every single person on the face the earth has one in their pocket. The effort to create stable diffusion is trivial when compared with many marvels of modern technology that we take for granted today. Stable diffusion in my pocket? It's already reality.
>There's been a lot of innovation in how you can best apply DNNs to problems, and that's where CLIP and stable diffusion innovate,
So? you talk as if this is a minor thing. It's very possible that enough innovations in this area and we'll find out that the entire human brain is layers and layers of complex "applications of DNNs". DNNs in theory can compute anything, so this path is forward is possible.
> but also a lot of the advancements has been purely due to sheer brute force from funding more compute resources to train deeper networks using larger datasets.
The training was trivial. Nobody expects AI technology to be able to be built by a single man. Like your iphone which lives in your pocket many lifetime hours of blood sweat and tears will need to be put into it. The half a million and 150,000 GPU hours is nothing. I expect more hours of training, larger datasets and eventually I expect all of tech to be encapsulated into some product that lives in the cloud or aka effectively in my pocket.
>I'm optimistic, but I'm also trying to contextualize to have the right expectations.
You don't realize it but this sentence is a declaration equivalent of saying "I'm a biased person." Optimism and pessimism is a form of bias. You state that you are optimistic, thus you stated you are biased. True logic has no leaning in either direction. This sentence here reveals what you are and that your "contextualization's" are inherently flawed.
Additionally, research finds that pessimistic people tend to be right. Yes logic dictates that neither pessimism or optimism is the way to go, but the data driven conclusion is that people who are pessimistic tend to be more realistic. The data shows that most humans are biased with an optimistic delusional bent.
This behavior is so pervasive that you even called yourself optimistic without realizing that it was a declaration of your own bias. Most reading your sentences miss it as well unconsciously thinking "optimism" of any form is more rational "pessimism" even though BOTH the logical and data driven conclusion show that this is not the case.
Contextualize yourself first before you contextualize the external world.
> You don't realize it but this sentence is a declaration equivalent of saying "I'm a biased person."
That was the point of my sentence lol, I'm explicitly stating my bias in order to address it.
Emotionally and for my own personal benefit, I want breakthroughs in AI to be achieved in my lifetime, and I hope they do.
I'm explicitly acknowledging and stating my bias for this. You address bias by recognizing it and taking it into account.
Then, I'm contextualizing my emotional stance, by providing data points about the reality of what the current state of AI is so that others have the context required so they can apply their own judgement.
Look, I get it, it's clear we're both excited about the prospects of AI, but you're getting emotionally defensive, that's exactly what I'm trying to avoid, let's remove our clout of excitement, and contextualize by looking at the data.
Yes we've made a lot of progress, but there's a lot more required, and unfortunately progress in any field of research has never followed a linearly constant progression, there was a AI winter before, there could be another.
The claim that we're close to having machines able to do everything that humans can do and do it even better doesn't seem to follow from the data.
There are gaps as I provided, we're only achieving close to human ability for tasks where we have hundreds of millions of examples. Even in those scenarios, there are catastrophic failure points in the models that humans would never fail at. Hardware wise, we're approaching the limits of how big a DNN we can train as well and how much data examples we have to feed it.
Models are getting really good at predictions over data, but they've yet to show any ability for creativity and invention. Right now they all need human assistance to output an interesting result. Take GPT-3, and it can't even interpret what you're saying, you've got to tailor how you talk to it to get interesting responses. Take stable diffusion, it hasn't been able to invent a new drawing style yet, only mimicking the ones it learns as another example. That's pretty major limitations compared to a human, and those are tasks that we have the most data available to us for training.
I think we'll need some other breakthrough to break past this barrier, and that I think could easily take much longer to achieve.
Your assessment could be different, maybe you take in all this data and predict a more optimistic timeline for overcoming all these issues, that's fair, what I wanted is to contextualize, not push my conclusion, contextualizing means providing others with the rationale and data so they can do their own assessment and arrive at their own conclusions.
There is a lot of optimism around Stable Diffusion and GPT-3 and others, you're flooded with the "OMG look at all it can do" data points, and then all the dreams of sci-fi becoming reality. There isn't as much exposure and discussions around, "okay but look at all it can't do and where the capabilities for what it does come from and how they were achieved". I wanted to provide that balance so people can have realistic expectations.
>okay but look at all it can't do and where the capabilities for what it does come from and how they were achieved
No. This is not the case. Everyone has your attitude. Every programmer is downplaying the significance of it all.
Literally look at the threads. Every tom dick and jane is saying it's not a big deal as if they know better. The truth is nobody knows shit. Your contextualization is just restating the obvious re-hashed opinion. "ML has limits it's not as good as the human brain it needs lots of training data, yadayadayada."
Your not contextualizing. Your reaffirming the prevailing bias.
> No. This is not the case. Everyone has your attitude. Every programmer is downplaying the significance of it all.
Expand on the "significance" part.
Because I don't feel like I'm downplaying the significance of it at all, this is revolutionary, and there's probably no stopping it, in 10 years you'll be using AI assistance for coding, writing, and art generation as a common thing, and it'll expand to other disciplines as well. It'll be everywhere in that regard. That could affect the job markets in a big way, this is a huge improvement to automation, either you'll need less artists, writers and programmers to get the same job done and so there will be less demand for that, or it'll lead to even more ambitious bigger projects and keep the demand for those jobs same but the output expectations will be higher.
Self-driving cars has some harder long tail to address to be fully trusted on its own, I think it could take more than 10 years for that, but assisted driving is already here and that'll just continue to get better.
I don't know about other programmers, but I don't feel like I'm downplaying the significance.
The limits I've mentioned are not made up, there's a lot we do know about them. There are theories of why DNN work like the information bottleneck, and we know more and more about them each year. We know now that you need to over parameterized, that deep layers are needed for generalization, we know that the only sizable improvements have been from even larger datasets over even larger networks. Whereas a child for lots of tasks only needs a single example to learn where models need thousands+.
Then we have things like the Winograd and Winogrande schemes, and we've observed that it seems the models get the right answers through statistical significance, not reasoning. Each time you adapt the question to remove the statistical significance they fall in accuracy again, until an even larger model is trained which likely just managed to capture even more statistical data points. So the problem of do the model actually understand the concept and can perform reasoning is a real one that currently has more sign of that they do not. Now it's possible that understanding concepts and performing logical reasoning as such is not as good as large statistical inference from giant corpus at performing certain tasks, and I think that's where you'll see models take over, for tasks where reasoning underperforms, or where it'll be an assisted tech where humans perform the reasoning and models the statistical inference.
This bit is key, there's still a possibility that models are actually learning concepts and reasoning, maybe if they just grow even larger they will for example. If that turned true, than it would be incredible, but it could also turn false, and models are just performing statistical inference, and what happens is we simply underestimated how good statistical inference can be at a lot of tasks to the point it can fool you in thinking they're understanding concepts and doing full human reasoning.
And you're right, we don't know for sure here, nobody knows for sure.
In what way do you feel I'm underestimating the significance here? Or what do you see as being "significant" that I haven't listed?
What I personally think is that maybe these AI are being really inefficient with their training and still capable of producing all of this wealth.
As we discover newer and better architectures, which I believe we will, then the machines are going to become better and better at doing this what they are doing.
So in other words, I think we already have enough data for most things. Certainly for art and code, writing, etc.
I think one issue is that the models learn by induction, not by deduction.
That means they learn the rules from the examples, but cannot reason about rules deductively.
Arguably, it means even if you were able to assemble the smallest amount of data that contained the entire array of scenarios, from which you'd learn the proper rules, it's unclear how small that would be, even assuming the most optimal learning model.
As I see it, we currently have good machine inductive learning, but are really far from any machine deductive learning.
Introducing causality into the model could very well just be another layer. In fact it is quite clear that stable diffusion produces a super set that is slightly bigger than solutions with causal reasoning taken into account. One more additional filter on top of that for causal deduction is a very realistic next step.
In a nutshell stable diffusion is like a calculator that gives me 3 answers. One of those answers is right, the other 2 are wrong. The next step in calculator creation is putting a filter over the output to get the right answer from the 3 possibilities.
>As I see it, we currently have good machine inductive learning, but are really far from any machine deductive learning.
Yeah, we already covered 50% of the AI equation in the past decade. The other deductive part is actively being researched and a revolution in your lifetime is HIGHLY probable given how fast the inductive part came to fruition.
There was a great exchange on Twitter the other day. Someone had generated anime style images of “Yann LeCun at Bell Labs working on convolutional neural networks." Yann LeCun responded with a comment that Totoro was next, then Meta AI replied with joke action items to generate storyboards with AI and set up a meeting with the studio.
Studio Ghibli responded with a picture of the wolf from Princess Mononoke, displeased.
I don't understand. Artists can also use these models to generate art, and looking at other industries, there are significant differences between professional output and "machine generated" output. Sort of the difference between a fullstack developer and no-code platform. Do developers fear no-code platforms taking over their jobs? Sure, yes for the menial tasks, but definitely not for high stakes complex products. Same as generative art - it may replace the need for those filler banners, stock photos, redrawing characters a little bit differently for that last panel - but no, you don't generate a coherent, entertaining 100 chapters manga with generative art.
Developers had a similar response where they don't think that training a model using their copyrighted work is fair use, especially for commercial use.
They'd expect that the model authors would need to comply with the open source license like for any other use.
This is basically the debate here, it's same for artists or developers.
The question is if it makes sense to allow someone to use your copyrighted work to train a model that they'd then use for commercial purposes, without needing a license agreement.
The debate over Copilot is not new either: at launch, it kept suggesting Carmack's fast square root code. (which got patched a day or so after). The recent shenanigans were about the code that was published under wrong license by third parties, and then picked up by Copilot.
Any overfitting (reproducing the training data verbatim) is not intended and is avoided as much as possible. But it's inevitable sometimes, and is a worse problem in code than art, as you have no way of knowing if it reproduced some private code verbatim.
And no, it's certainly not the entire debate, many people are also afraid that their job will be automated. Which is a silly assumption in case of Copilot as well as with image gen models - it stems from the misunderstanding of what it can and can't do, what's fundamentally possible with it in the future etc.
The moral panic is strong, but I have the feeling that it's amplified by social media without really being substantiated. The advent of CGI was the same, it just was more quiet due to the lack of social media at the time.
Unless you are really good, a big name, or willing to draw porn, it's quite difficult to make a living with art. Fortunately, AI doesn't quite know how to draw penetration yet -- and even when it learns, the big money is in furry porn so it will have to learn again how to draw convincing animal-people engaged in convincing coitus.
Systems like Copilot and Dall-E and so on turn their training data into anonymous common property. Your work becomes my work. This may appeal to naive people (students, hippies, etc.), for whom socialist/communist ideas are attractive, but it's poison in the real world.
These systems are a mechanism that can regurgitate (digest, remix, emit) without attribution all of the world's open code and all of the world's art.
With these systems, you're giving everyone the ability to plagiarize everything, effortlessly and unknowingly. No skill, no effort, no time required. No awareness of the sources of the derivative work.
My work is now your work. Everyone and his 10-year old brother can "write" my code (and derivatives), without ever knowing I wrote it, without ever knowing I existed. Everyone can use my hard work, regurgitated anonymously, stripped of all credit, stripped of all attribution, stripped of all identity and ancestry and citation.
It's a new kind of use not known (or imagined?) when the copyright laws were written.
Training must be opt in, not opt out.
Every artist, every creative individual, must EXPLICITLY OPT IN to having their hard work regurgitated anonymously by Copilot or Dall-E or whatever.
If you want to donate your code or your painting or your music so it can easily be "written" or "painted", in whole or in part, by everyone else, without attribution, then go ahead and opt in.
But if an author or artist does not EXPLICITLY OPT IN, you can't use their creative work to train these systems.
All these code/art washing systems, that absorb and mix and regurgitate the hard work of creative people must be strictly opt in.
If you don't care about this, it's naivete, or a lack of foresight, or apathy as these companies pillage the commons. Not something to be proud of.
Microsoft and OpenAI (and others) are robbing us and you should care.
So the year is 2120. We could have tech that allows anyone to generate code and art "effortlessly and unknowingly", but we don't because we don't want to upset some dead folks from a hundred years ago. Is that the society we want?
so as long as in 2120 the tech is trained on public domain artists and code, it will be ok.
But if it is trained copyrighted sources, AND requires its own attributions without actually requiring the copyrighted source attributions and/or its licenses, then you are right, we won't get it. The same way we have SSPL and other licenses in addition to public domain code.
if you reject copyright (I assume that is what you mean by 'having nice things') it has to work both ways and not the"Disney I love Public Domain but I lobby to extend copyright forever" way, because it suits whatever business model.
Pretty much everyone can down and use the publicly available models. And for those without the hardware, there are cheaply available online services you can use as well.
You can use it as much as everyone else. Problem solved.
I don't seem to understand why you think the access to these tools is a bad thing.
Like, I can deduce that your argument comes from a place of "this work is mine and mine alone"... but is it? We don't have copyright on mathematics for example, if you come up with an equation or a solution to a tricky problem I can make use of it without the need of attribution. And in many ways, attribution can work similarly in AI generated art; it is common for people seeking a distinctive artists style to write prompts that include said artist and the AI, in turn, will willfully comply.
But what's the real core of your argument, why is it that you think information that is shared in this way is bad?
I can actually empathize with artists that think this technology might threaten their livelihood but this particular position I also view as overly-catastrophic. But you don't really seem to be concerned with that and instead you're making an argument that comes solely from individuality? But to me, these AI models are the maximum expression of something that as societies we have been doing for a very long time: to take the work that's available and use it to create something new.
Now, I completely agree with the licensing issue when it comes to github's copilot. It used public code to train, and therefor any work that is produced through it should be public as well. It's a different issue.
But your argument that "your work becomes my work" is bad is not something I fully understand. Why is it bad? Why is it poison? Shouldn't we maybe be questioning the system that envelops us if such a thing as sharing is considered poison?
> Now, I completely agree with the licensing issue when it comes to github's copilot. It used public code to train, and therefor any work that is produced through it should be public as well. It's a different issue.
Artwork can be very much licensed the same way that code can. If I train my AI on Creative Commons Share Alike or None Commercial licenses then why should that be any different?
And if we honour one sort of licensing, the copyleft side, why should we treat the other side, the copyright differently?
Either AI is allowed to ignore copyright and licenses for everything or it's not, copy or painting, it makes no difference.
I, as an consumer, don't care if anime/manga/textures/etc... are created via hand or AI.
I just want a good story or good entertainment.
Thus, my vote with money will go to whoever creates the best story or entertainment.
I don't care about you needing 20 hours to create a handpainted manga, I want my entertainment: and if someone who cannot draw uses AI to create a 10x better story than someone who can paint but is shit at stories...well..though luck, I don't care.
Probably true to some extent, but that's also why copyright laws are put in place, so there could definitely be laws made that could require licensing if you use it to generate an AI and then use it for commercial purposes.
The time-window to have the debate of if you should have such laws or not is now.
The internet is global. Idiosyncratic copyright laws tend to just get ignored. It’s illegal (in France) to post a photo of the Eiffel Tower at night, but if you search “Eiffel tower at night” on Google images, you will find plenty of examples. EU database rights are a similarly toothless concept and are just ignored most of the time.
If someone sets up an AI anime shop in Mozambique, who is going to stop them?
Practically speaking, and the way things are going, I think you're absolutely right.
But theoretically, countries can ban your product if it doesn't comply with local copyright laws. That can mean that say any video game, movie or tv show made with AI generated models that break copyright laws in say US could be banned from being sold in the US.
You can even go a step further, and sanction the company, blocking all payment processors that are looking to process payment in the US from disbursing money to that company.
So in practice I think you're right, and I'd advise that artists, developers and everyone else braces themselves to it and find ways to stay marketable and relevant even in the age of AI, but in theory it's just a political thing, and if there was real will to do something about it, it could be done.
That's not how it works. AI is copying/stealing from people who actually create. "Laundered legal liability" as someone here so eloquently put it. When co-pilot reproduces GPL code verbatim, it's not AI, it's just convoluted copyright infringement. I have no doubts this is happening with anime too. The so called AI is just a thief. It's not creating new works, it's copying parts of works and making a composite.
When I write, I am doing nothing but laundering sentences and grammar rules from my English textbooks as well as sampling from the millions of sentences that I have read in my lifetime to compose into this comment as a response to your prompt. When I code, I heavily reference existing implementations and often only make minor stylistic changes, copying algorithms and data structures wholesale from memory. When I cook, I'm literally copying the contents of the cookbooks or cooking channels with a few minor substitutions at best.
Unfortunately, I am not brilliant enough to come up with my own algorithms, recipes, or literary flourishes. Few people are truly "creators", to be honest, and even creators build off or derive from existing work.
It doesn't copy verbatim, neither does it copy parts. Sure, if you request "Sunflowers by Van Gogh" you will get something very close to Sunflowers by Van Gogh. That's because Van Gogh produced several similar works with sunflowers, because models are notoriously overfit on those, but more importantly because you asked for it specifically; you could have asked Google and get unlimited copies of Sunflowers that are much more precise. Good luck accidentally describing Sunflowers in abstract terms, though. So, you would be responsible for plagiarism if you publish a work of art created by someone while claiming it's yours, not the model creators.
I think there is a separation between aesthetic and artistic value. Mona Lisa copies have the same aesthetic but different artistic values. I think folks who care are going to differentiate on that.
edit: also, see Duchamp's Fountain
Duchamp's Fountain is a great example, because most people couldn't care less about its artistic value because they reject it for its lack of aesthetic value. Just like you don't see people gluing themselves to postmodern art which emphasizes what is artistic over what is aesthetic.
Most people don't care about what is artistic. It's the experience that matters.
AI art will be something different, and because it's so cheap it'll eradicate everything else (at least most of its mainstream exposure)
That's what's interesting, to understand what we're losing and what we're gaining.
I believe art is an integral part of our humanity, this radical change in art, will definitely cause a radical change in what it is to be human.
I personally tend to think it'll take current capitalistic trends to the extreme. Automatically personalized optimized art for purely hedonistic entertainment. This will be what we'll mostly experience.
We'll lose that fragile bridge between the islands that are our individualities that art allows. The transmission of internal mental process of one person, her unique one of a kind perception and expression experienced to some degree by another. As a way to communicate, not just to cause pleasure.
It's a huge loss IMHO. It's much more substantial than copyright issues and jobs. The artists positions in the article articulate this, for some reason quotes like: “Artists are not just a ‘style.’ They’re not a product. They’re a breathing, experiencing person,” are being read as artists being scared for their livelihood. Why not read it as it's written? As artists fearing for their humanhood?
Was it? I'd put it at almost all negative reactions to copilot being about its tendency to reproduce chunks of code verbatim in a way that makes it look likely to be copyright infringement. In contrast, my impression is that the art folks are more upset at the being-automated-out-of-a-job angle. Not universally true, of course, but those really are different concerns.
No it wasn't. There were many opinions, just as there are amongst artists, including a small number of very loud negative voices.
But I'm pretty sure the reaction of most "geeks" was positive. Geeks generally love new tech, especially when it can improve their workflow.
A very valid negative criticism is that a huge company with a bad track record having ownership of a for-profit tool created by processing the work of millions of people is not good.
However that criticism shouldn't be conflated with criticizing the existence of the tool itself.
71 comments
[ 1.8 ms ] story [ 140 ms ] threadThat said, maybe we don't train models on living artists or watermarked images, ok? It just splashes it with turds and vomit, IMHO. Like what happened to Cleverbot. Gunked up with trolls and idiots, and technically non-consensual too. Maybe we stop, eh? So tasteless.
This one example? Fine. Guy's dead, it's a dedication. Whatever.
We need UBI/social safety/cultural reform to support artists in doing art, a thing that's deeply important. We don't need IP laws that allow people to stifle creativity in order to collect rents on things that can be infinitely replicated for free.
Which do you put in place first?
I think this is a bit of the issue with this argument. As automation does eat at jobs, you can say that the solution is UBI or something like it, but since that's not there, right now and until it happens (which could be never), it will affect people's livelihood, and it's fair for them to be pushing back.
If such social foundation was already here, I don't think people would be complaining as much, they'd probably be joyful that they'd no longer need to work as hard.
If artists cared enough, they need to make an union of some sort and lobby for non synthetic art labelling for protection like big Diary and get AI generated art to be branded like GMO foods.
I feel bad for anime artists (well all artists), they work particularly hard. But there are a lot of good manga drawn by mediocre artists, if there's option to change them stylistically closer to much more proficient masters (like a video game remaster), I'd do it in a heart beat. There's going to be audience out there for OG Tomohirio One Punch Man squiggles, but there's a reason it gained mainstream popularity after being redrawn by Murata.
I'm also just waiting for AI to generate music based off the few particular songs I like.
Now for the first time in my life I can see the future beyond science fiction. Where machines are capable of doing everything humans do, and do it better.
The outrage is a violent attempt to stop the inevitable. I sympathize. Because artists won't be the only victims of this change when it's all over.
Sure hope so. Then I can move even closer to self-actualization than I am now.
If there had ever been a single inflection point in history where Luddism was the right answer in the long run, I'd be more sympathetic to the pessimistic point of view. As things stand, though, doesn't their argument amount to a special pleading fallacy? "It's different this time!" No... no, it's not.
>Sure hope so. Then I can move even closer to self-actualization than I am now.
Yeah and then you starve to death because you have some measly salary. Who says you're the owner of the robot who took over your job? It may be possible, but thinking this is the only possibility is not intelligent. Every realistic possibility must be considered.
With wealth inequality as it is right now, a Few plutocrats reaping most of the benefits of AI with the majority of people scraping by is not just the more realistic possibility, it is also remarkably similar to the current reality.
>If there had ever been a single inflection point in history where Luddism was the right answer in the long run, I'd be more sympathetic to the pessimistic point of view. As things stand, though, doesn't their argument amount to a special pleading fallacy? "It's different this time!" No... no, it's not.
Oh there are tons of examples. You just view history through a biased lens. You're unable to see the times when technology did harm the population. I utterly guarantee that your conclusion is not data driven. You simply are unconsciously picking and choosing the data to support your optimism.
A big portion of the wealth inequality we see today is the result of a century of technological advancement. But you're partly blind to it because you're on the positive side of the wealth inequality equation.
None of which you seem to be able to cite.
As for "wealth inequality," because of capitalism, our homeless people live better than kings and princes did a few hundred years ago. I'm OK with that.
Able? The examples are obvious. I can cite them though. Once I cite them you should should 180 degree your opinion. But of course you won't. You're so biased you can't even see the obvious examples all around you:
https://www.4ocean.com/blogs/blog/a-brief-history-of-the-pla... The plastic crisis. The result of the invention of plastic.
https://www.clientearth.org/latest/latest-updates/stories/fo... The discovery of how to extract energy from fossil fuels looks to potentially herald massive damage to society and already has effected the lives of many.
https://rainforests.mongabay.com/09easter_island.htm Actual civilization destroyed through the introduction of wood based technology to an island resulting in deforestation and complete collapse.
https://web.colby.edu/st297-global18/2018/10/30/has-technolo... Obesity epidemic caused by ultra processed food technology. Industrialization and the negative effects. "At least they're not starving and living better than kings" is what I imagine coming out of your privileged mind.
https://www.europarl.europa.eu/RegData/etudes/IDAN/2018/6190... Fake news, misinformation created by social media technology. Before free press and technology people were simply less informed. Now, the current generation of people are likely the most informed but also the most misinformed people who've ever lived. People think they live in a more dangerous world, but in reality crime has gone down. This is all because of the current news cycle.
https://www.medicalnewstoday.com/articles/negative-effects-o... health issues with usage of internet related technologies.
https://www.icanw.org/catastrophic_harm Nuclear weapons. The discovery of the relationship between energy and mass and other advancements in physics and chemistry led to the development of technology that enables humanity to literally destroy the world if they chose to. This example literally expands the spectrum about the amount of harm technology can do. In less then a day ICBM and nuclear bomb technology can level the world.
https://en.wikipedia.org/wiki/Chernobyl_disaster Nuclear power melt downs. There are several. This is the most famous one. Obviously there's nuance to this and I will address it. The net benefit of nuclear power is better then fossil fuels in terms of death and destruction when taken from a statistical standpoint. But you still cannot ignore the fact that a single mistake leveled a whole town. Blindly charging forward because you trust the numbers is equally ignorant.
Need I go on? There are OBVIOUS benefits of technology I would be blind to not to see it. But I'm also able to see the OBVIOUS negative side as well. The nuance.
I would be wrong if I said you didn't know about each of these things. You know it. These t...
Fossil fuels were not an example where Luddites were right. They have lifted billions out of poverty over hundreds of years at a comparatively small cost, to say nothing of saving countless millions from freezing to death in the dark. Now it's time to leave them behind, and that's just what we're doing.
Rain forest destruction is not an example where Luddites were right. I don't even know where you're coming from here. It should be stopped, but the people saying it should be stopped aren't Luddites.
Obesity and misinformation are not examples where Luddites were right. Again, you're just flailing here, pulling random crap out of your ass. Modern farming practice has saved billions (ever heard of a fellow named Borlaug? It's Google time.) And despite the growing presence of extremism caused by bad actors taking advantage of the same technology that's available to the rest of us, the average person on the street is better informed than ever. If they aren't, it's not the media's fault, it's theirs alone.
"Negative effects of the Internet" reported on an Internet quack medical news site. Okey-dokey, then.
Nuclear weapons put a stop to increasingly-destructive world wars by doing the unthinkable: making the leaders fear for their own lives. That's a boon unmatched by any other in post-medieval history. And nuclear technology could have saved us from countless other ills, if it weren't for mindless FUD from celebrities and politicians with no sense of risk assessment.
Chernobyl? See above. Fossil fuel power sources have killed unimaginable magnitudes more people than nuclear power plants ever could, which is one reason why we need to move past them. But they do it in boring, remote places like hospitals and cancer wards, and they do it years after the harm is inflicted. Without scary green blobs of molten radioactive crap to show on TV, nobody including you cares. News flash: faulty reasoning leads to poor outcomes, (no) film at 11.
Whether you should "go on" or not is up to you. You appear to be addressing a question that wasn't asked, or that was asked by someone who isn't here.
>The "plastic crisis" is not an example where Luddites were right. At the time it seemed like a good idea to make everything out of plastic, and it probably was...
"Seemed like a good idea"? and "probably right"? Wow. you have no confidence in what you're talking about here. Your language implies that you AREN'T even sure about the hypothetical alternative if we didn't take the plastic route. The rest of the argument is basically just a rant on how to "solve" the problem and in no way a counter to the fact that plastic technology has harmed the planet. You ingest a credit card worth of plastic per week thanks to the "plastic crisis".
>Fossil fuels were not an example where Luddites were right. They have lifted billions out of poverty over hundreds of years at a comparatively small cost, to say nothing of saving countless millions from freezing to death in the dark. Now it's time to leave them behind, and that's just what we're doing.
More hypothetical's. Nobody was lifted out of poverty. What happens is that food production increased so the planet can support a bigger population. Without that production increase those people wouldn't have existed in the first place. Society existed fine at lower pre-industrial state with a much lower and sustainable population just fine without fossil fuels.
This discussion of hypotheticals is a form of rationalization. There is literally no denying that each of the technologies I list here contributes a negative to society.
>Obesity and misinformation are not examples where Luddites were right. Again, you're just flailing here, pulling random crap out of your ass.
No. Nothing's pulled out of my ass. It's just you're unable to comprehend anything. Let me spell out to you what I'm talking about. I'm talking about INDUSTRIAL FOOD PROCESSING. Not industrial farming. Farming doesn't cause obesity. Food processing does when you strip a food product down to levels of unnatural purity. Go read it again. You're such a genius. Unbelievable.
>"Negative effects of the Internet" reported on an Internet quack medical news site. Okey-dokey, then.
Bro. Everybody on the face of the earth knows the negative psychological effects of social media. The negative health effects of people spending too much time gaming on the internet. Everybody knows this. There's a million sites talking about this. It's OBVIOUS. You want a "better" source? Here: https://web.stanford.edu/~gentzkow/research/facebook.pdf random research paper from stanford. But this is besides the point. You're just arguing to stay alive right now.
I simply "cited" a source to show you how it CAN be cited. But literally you pop up here laughing at the source? It's like I proved to you with a dictionary that O-N-E spells ONE and you just laugh at my source and don't even believe that O-N-E spells one.
Wow just wow.
>Nuclear weapons put a stop to increasingly-destructive world wars by doing the unthinkable: making the leaders fear for their own lives. That's a boon unmatched by any other in post-medieval history. And nuclear technology could have saved us from countless other ills, if it weren't for mindless FUD from celebrities and politicians with no sense of risk assessment.
This is next level mind twisting. The amount of bias you have to have in order to prove your point must be so high that you can even justify nuclear weapons as good. You're too far gone man. Leaders fearing for their own lives as if leaders were never ever under any sort of threat to their lives befor...
That's not really what this site is for. Your talents would be better put to use over on Reddit, or maybe Fark.
No. You don't get it. It's literally impossible to change a mans opinion. Everyone is too biased and they have their pride tied up with their arguments. Dispassionate logic is a fantasy.
Changing someones opinion is a sign of an effective discussion. Which is basically what NEVER happens on HN. HN is just people patting each other on the back while never effectively conveying information or changing someones opinion. If your opinion changed or even if you got defensive around your position it means you learned something. The discussion was effective and Informative.
That is the purpose of HN. It's just too bad being informed that you're utterly wrong feels bad and is seen as a bad thing. It shouldn't be this way. Realizing you're wrong should be a form of enlightenment.
Also it's not a skill. Being informative is simply about being MORE correct and right then the other party and Informing them about it. Which is what happened to you.
I think there is a big difference between feeding an AI model a wide swath of an art style like anime and creating and using a model.
Taking one persons particular work and style and copying in competition or near competition with that person is something that I don't think is inevitable and something I would be happy to see the expansion of legal protection for.
I know I will get downvoted since HN hates legal protection for artists, but I just don't see the societal benefit from allowing an individual human to be replicated by AI without their consent.
If AI art is better than what humans can make then this is more than a simple moral issue.... basically the inevitable future is that the entire occupation of "art" will be steam rolled and 99% of what's out there will become computer generated. Morals be damned, it won't even be a factor. A bunch of scholarly types trying to quantify the rightness and wrongness of something isn't going to mean jack when the power to do create art is so easy and trivial.
In fact the morals will shift with the culture and technology and it will be a violent shift. People getting angry here is just the beginning of bigger conflicts in the future. If AI keeps improving things will stabilize into an age of humanity where original art is but a cheap commodity, but that age is not one that will arrive soon, the step before that is violence.
The even scarier part is that it's not just art that's being targeted here. Art is just one of the first victims.
In the end it allowed more wealth and created more demand for music. There will be more demand for art.
More demand for art is like saying more demand for air, more demand for oxygen. Demand is saturated and so is supply. What's happening here is similar to outsourcing. Jobs are going to another place, just not a real place and it isn't people replacing your job.
Your analogy is biased. You picked a specific example of new technology being "not bad". How about picking an example of technology being bad? Plenty of analogies in the opposite direction. Singular examples and analogies are not proof of anything.
Anyway I think the scare about AI generated art is misplaced when it comes to art as a general concept. It is after all merely a tool and a tool can suppress art. Plenty of abstract art was already not limited to technical mastery and could have been made by a computer yet abstract art didn’t disappear and artists weren’t displaced.
The Japanese and Korean illustrators scare however is not misplaced. It comes from how the industry is structured. For every credited artists or directors, you have a whole team of assistants and illustrators who have a technical role and produce parts of the image imitating the desired style. This, I could see replace by AI and as it is the entry level position for a career in the field, it would significantly modify its functioning. These people are scared for their livelihood and they probably are right to be.
That's not demand for art. That's demand for learning to become an artist. It's like saying people wanting to become farmers is equivalent to more demand for food. No. Related? Yes. But they are not the same.
>Anyway I think the scare about AI generated art is misplaced when it comes to art as a general concept. It is after all merely a tool and a tool can suppress art. Plenty of abstract art was already not limited to technical mastery and could have been made by a computer yet abstract art didn’t disappear and artists weren’t displaced.
You can think this all you want. But you can also project the current trends of technology into the future and envision a future where AI creates art better and faster then ANY human on the face of the earth. Then take that AI and apply it to programming, decision making, car driving and every other occupation that right now only humans can do.
Notice how I just made a realistic projection? I didn't say whether it was scary or horrible. I just made a realistic projection and THAT projection itself should give ANYONE pause about the future of humanity. You are way to optimistic. I leave it to you to fill in what the projection means for us.
A lot of people are just like "Oh we'll just switch to UBI, and AI will create a utopia". Yeah, that's not a realistic projection. Unlike my projection, this is just a random shot in the dark. Given that utopia and UBI never successfully existed on such a massive scale yet, such a transition should also give you pause. Additionally UBI is dangerously similar to communism which is in itself questionable.
>The Japanese and Korean illustrators scare however is not misplaced. It comes from how the industry is structured
No it's from how reality is structured. If some machine does your job better than you then you become redundant. No amount of industrial structuring can make this FACT not true.
>These people are scared for their livelihood and they probably are right to be.
You should be scared as well. AI targets almost every job. Art is just the first to fall.
Well, you haven’t. There is absolutely nothing in the current AI state of the art which points towards that. As many as been telling you, you are scared of ghosts.
Did the loop make workers disappear? Did cameras made us stop painting?
What you wrote were written in exactly the same way for all these innovations. You will pass, progress will march on.
It's called logic, there are no ghosts here. If improvements arrive incrementally as they have been for the last decade with milestones like Dall-E, GPT-3, Alpha Go, Deep Blue. These events form what's called an "upward trend". If you project that trend following mathematical statistical procedures with a trend line, you can "project" that line into the future. That is called a "linear projection".
That trend is far more mathematical, far more logical then any other evaluation you can come up with. Everything you have are just qualitative opinions. Literally I have time on the X-axis and "amount of things computers are able to do almost as good if not better then humans" on the Y-axis. That trend line flows back from, the inception of computing. Not just a decade of data but half a century. My conclusions are DATA driven. Your conclusions are just gut feelings.
If you deny this you are illogical. There is no other way of putting this.
>Did the loop make workers disappear? Did cameras made us stop painting?
Wtf is the loop? Did automation in factories make workers disappear? Did cameras make us stop painting?
The answer is yes. Many painters who use to go to peoples houses to paint portraits.... most of those occupations HAVE disappeared. Same with factory workers, automation wiped a lot of them out (aka disappeared). You can see the results of it today... the wealth inequality we see in the world is partly the result of technology.
You are in huge denial if you think technology doesn't make occupations disappear. This kind of thing is a Well known economic phenomenon NOT exclusive to AI. When technology improves workers in related fields become less. There are loads and loads of quantitative line graphs with trend lines that show this and economists who can back it up.
>What you wrote were written in exactly the same way for all these innovations. You will pass, progress will march on.
What do you mean I will pass? Of course progress will march on. There's NOTHING anyone can do to stop it. AI will steam roll entire industries including eventually itself. You already have ML building models and ML training itself.
You're like a climate denier. I'm saying global warming is inevitable. It will happen. There is nothing left humanity can do to change the current trend. You're saying there is no trend. No evidence. No data within the last century at all. Not to be insulting but this is total and complete blindness. Literally I am describing to you trend lines and data points and you're unable to see how biased you are.
Yeah, just think of all those farmers who had no choice but to commit suicide when threshers, steam engines, and milking machines were invented.
You're confusing a boon with a tragedy. It's a GOOD thing to be made redundant by automation. Now you can do what you were supposed to be doing all along... whatever that is.
Yeah. Many farmers did become redundant. You're not thinking clearly.
>You're confusing a boon with a tragedy. It's a GOOD thing to be made redundant by automation.
You're an extremist thinking all technological advancement is a boon never considering the nuances of situations. It can be good and it can be bad. Obesity for example is the result of improved efficiencies in automated manufacturing of food.
> Now you can do what you were supposed to be doing all along... whatever that is.
What's funny is you don't even know what they're suppose to be doing. What's a human being suppose to do when every ounce of his capability is rendered redundant by machines? What happens to the economy how does it work? You don't know and you think it's a good thing. That's bias right there. Not knowing something and thinking it's good AS IF you know it. You can't know something is good without knowing what that something is.
You know how the economy works right? Let me tell you what it actually is. It's an UNKNOWN thing. Never in the history of the human race has this type of economy existed where most humans no longer need to be productive. If you think everyone is going to have equal access to the benefits of AI and automation history says this isn't what typically happens.
Maybe one day, but we're still far away. To contextualize, Stable Diffusion needed 5.83 billion images to be trained on, and it also uses another model CLIP that needed 400 million labeled images to be trained.
That means that all tasks where we don't have hundred of millions or more of available accumulated datasets are really far off from being learnable by current ML models.
But that's just about stable diffusion. You need to measure the speed of how fast stable diffusion came to be. 3 years ago all of this stuff was thought to be impossible. Now it's a SAS. What does the next 3 years bring? Contextualize that.
Nobody knows when that day will come, BUT within our lifetimes is a very realistic bet given the speed in which the ML industry is changing and growing.
Also, Stable Diffusion isn't a revolution in model architecture, it's still a deep neural network. There's been a lot of innovation in how you can best apply DNNs to problems, and that's where CLIP and stable diffusion innovate, but also a lot of the advancements has been purely due to sheer brute force from funding more compute resources to train deeper networks using larger datasets.
I'm optimistic, but I'm also trying to contextualize to have the right expectations.
So? The iphone as it is today took EVEN more man hours and billions of dollars to create. Yet every single person on the face the earth has one in their pocket. The effort to create stable diffusion is trivial when compared with many marvels of modern technology that we take for granted today. Stable diffusion in my pocket? It's already reality.
>There's been a lot of innovation in how you can best apply DNNs to problems, and that's where CLIP and stable diffusion innovate,
So? you talk as if this is a minor thing. It's very possible that enough innovations in this area and we'll find out that the entire human brain is layers and layers of complex "applications of DNNs". DNNs in theory can compute anything, so this path is forward is possible.
> but also a lot of the advancements has been purely due to sheer brute force from funding more compute resources to train deeper networks using larger datasets.
The training was trivial. Nobody expects AI technology to be able to be built by a single man. Like your iphone which lives in your pocket many lifetime hours of blood sweat and tears will need to be put into it. The half a million and 150,000 GPU hours is nothing. I expect more hours of training, larger datasets and eventually I expect all of tech to be encapsulated into some product that lives in the cloud or aka effectively in my pocket.
>I'm optimistic, but I'm also trying to contextualize to have the right expectations.
You don't realize it but this sentence is a declaration equivalent of saying "I'm a biased person." Optimism and pessimism is a form of bias. You state that you are optimistic, thus you stated you are biased. True logic has no leaning in either direction. This sentence here reveals what you are and that your "contextualization's" are inherently flawed.
Additionally, research finds that pessimistic people tend to be right. Yes logic dictates that neither pessimism or optimism is the way to go, but the data driven conclusion is that people who are pessimistic tend to be more realistic. The data shows that most humans are biased with an optimistic delusional bent.
This behavior is so pervasive that you even called yourself optimistic without realizing that it was a declaration of your own bias. Most reading your sentences miss it as well unconsciously thinking "optimism" of any form is more rational "pessimism" even though BOTH the logical and data driven conclusion show that this is not the case.
Contextualize yourself first before you contextualize the external world.
That was the point of my sentence lol, I'm explicitly stating my bias in order to address it.
Emotionally and for my own personal benefit, I want breakthroughs in AI to be achieved in my lifetime, and I hope they do.
I'm explicitly acknowledging and stating my bias for this. You address bias by recognizing it and taking it into account.
Then, I'm contextualizing my emotional stance, by providing data points about the reality of what the current state of AI is so that others have the context required so they can apply their own judgement.
Look, I get it, it's clear we're both excited about the prospects of AI, but you're getting emotionally defensive, that's exactly what I'm trying to avoid, let's remove our clout of excitement, and contextualize by looking at the data.
Yes we've made a lot of progress, but there's a lot more required, and unfortunately progress in any field of research has never followed a linearly constant progression, there was a AI winter before, there could be another.
The claim that we're close to having machines able to do everything that humans can do and do it even better doesn't seem to follow from the data.
There are gaps as I provided, we're only achieving close to human ability for tasks where we have hundreds of millions of examples. Even in those scenarios, there are catastrophic failure points in the models that humans would never fail at. Hardware wise, we're approaching the limits of how big a DNN we can train as well and how much data examples we have to feed it.
Models are getting really good at predictions over data, but they've yet to show any ability for creativity and invention. Right now they all need human assistance to output an interesting result. Take GPT-3, and it can't even interpret what you're saying, you've got to tailor how you talk to it to get interesting responses. Take stable diffusion, it hasn't been able to invent a new drawing style yet, only mimicking the ones it learns as another example. That's pretty major limitations compared to a human, and those are tasks that we have the most data available to us for training.
I think we'll need some other breakthrough to break past this barrier, and that I think could easily take much longer to achieve.
Your assessment could be different, maybe you take in all this data and predict a more optimistic timeline for overcoming all these issues, that's fair, what I wanted is to contextualize, not push my conclusion, contextualizing means providing others with the rationale and data so they can do their own assessment and arrive at their own conclusions.
There is a lot of optimism around Stable Diffusion and GPT-3 and others, you're flooded with the "OMG look at all it can do" data points, and then all the dreams of sci-fi becoming reality. There isn't as much exposure and discussions around, "okay but look at all it can't do and where the capabilities for what it does come from and how they were achieved". I wanted to provide that balance so people can have realistic expectations.
No. This is not the case. Everyone has your attitude. Every programmer is downplaying the significance of it all.
Literally look at the threads. Every tom dick and jane is saying it's not a big deal as if they know better. The truth is nobody knows shit. Your contextualization is just restating the obvious re-hashed opinion. "ML has limits it's not as good as the human brain it needs lots of training data, yadayadayada."
Your not contextualizing. Your reaffirming the prevailing bias.
Expand on the "significance" part.
Because I don't feel like I'm downplaying the significance of it at all, this is revolutionary, and there's probably no stopping it, in 10 years you'll be using AI assistance for coding, writing, and art generation as a common thing, and it'll expand to other disciplines as well. It'll be everywhere in that regard. That could affect the job markets in a big way, this is a huge improvement to automation, either you'll need less artists, writers and programmers to get the same job done and so there will be less demand for that, or it'll lead to even more ambitious bigger projects and keep the demand for those jobs same but the output expectations will be higher.
Self-driving cars has some harder long tail to address to be fully trusted on its own, I think it could take more than 10 years for that, but assisted driving is already here and that'll just continue to get better.
I don't know about other programmers, but I don't feel like I'm downplaying the significance.
The limits I've mentioned are not made up, there's a lot we do know about them. There are theories of why DNN work like the information bottleneck, and we know more and more about them each year. We know now that you need to over parameterized, that deep layers are needed for generalization, we know that the only sizable improvements have been from even larger datasets over even larger networks. Whereas a child for lots of tasks only needs a single example to learn where models need thousands+.
Then we have things like the Winograd and Winogrande schemes, and we've observed that it seems the models get the right answers through statistical significance, not reasoning. Each time you adapt the question to remove the statistical significance they fall in accuracy again, until an even larger model is trained which likely just managed to capture even more statistical data points. So the problem of do the model actually understand the concept and can perform reasoning is a real one that currently has more sign of that they do not. Now it's possible that understanding concepts and performing logical reasoning as such is not as good as large statistical inference from giant corpus at performing certain tasks, and I think that's where you'll see models take over, for tasks where reasoning underperforms, or where it'll be an assisted tech where humans perform the reasoning and models the statistical inference.
This bit is key, there's still a possibility that models are actually learning concepts and reasoning, maybe if they just grow even larger they will for example. If that turned true, than it would be incredible, but it could also turn false, and models are just performing statistical inference, and what happens is we simply underestimated how good statistical inference can be at a lot of tasks to the point it can fool you in thinking they're understanding concepts and doing full human reasoning.
And you're right, we don't know for sure here, nobody knows for sure.
In what way do you feel I'm underestimating the significance here? Or what do you see as being "significant" that I haven't listed?
As we discover newer and better architectures, which I believe we will, then the machines are going to become better and better at doing this what they are doing.
So in other words, I think we already have enough data for most things. Certainly for art and code, writing, etc.
That means they learn the rules from the examples, but cannot reason about rules deductively.
Arguably, it means even if you were able to assemble the smallest amount of data that contained the entire array of scenarios, from which you'd learn the proper rules, it's unclear how small that would be, even assuming the most optimal learning model.
As I see it, we currently have good machine inductive learning, but are really far from any machine deductive learning.
In a nutshell stable diffusion is like a calculator that gives me 3 answers. One of those answers is right, the other 2 are wrong. The next step in calculator creation is putting a filter over the output to get the right answer from the 3 possibilities.
>As I see it, we currently have good machine inductive learning, but are really far from any machine deductive learning.
Yeah, we already covered 50% of the AI equation in the past decade. The other deductive part is actively being researched and a revolution in your lifetime is HIGHLY probable given how fast the inductive part came to fruition.
Studio Ghibli responded with a picture of the wolf from Princess Mononoke, displeased.
https://twitter.com/JP_GHIBLI/status/1573584657668444161?cxt...
Developers had a similar response where they don't think that training a model using their copyrighted work is fair use, especially for commercial use.
They'd expect that the model authors would need to comply with the open source license like for any other use.
This is basically the debate here, it's same for artists or developers.
The question is if it makes sense to allow someone to use your copyrighted work to train a model that they'd then use for commercial purposes, without needing a license agreement.
Any overfitting (reproducing the training data verbatim) is not intended and is avoided as much as possible. But it's inevitable sometimes, and is a worse problem in code than art, as you have no way of knowing if it reproduced some private code verbatim.
And no, it's certainly not the entire debate, many people are also afraid that their job will be automated. Which is a silly assumption in case of Copilot as well as with image gen models - it stems from the misunderstanding of what it can and can't do, what's fundamentally possible with it in the future etc.
The moral panic is strong, but I have the feeling that it's amplified by social media without really being substantiated. The advent of CGI was the same, it just was more quiet due to the lack of social media at the time.
These systems are a mechanism that can regurgitate (digest, remix, emit) without attribution all of the world's open code and all of the world's art.
With these systems, you're giving everyone the ability to plagiarize everything, effortlessly and unknowingly. No skill, no effort, no time required. No awareness of the sources of the derivative work.
My work is now your work. Everyone and his 10-year old brother can "write" my code (and derivatives), without ever knowing I wrote it, without ever knowing I existed. Everyone can use my hard work, regurgitated anonymously, stripped of all credit, stripped of all attribution, stripped of all identity and ancestry and citation.
It's a new kind of use not known (or imagined?) when the copyright laws were written.
Training must be opt in, not opt out.
Every artist, every creative individual, must EXPLICITLY OPT IN to having their hard work regurgitated anonymously by Copilot or Dall-E or whatever.
If you want to donate your code or your painting or your music so it can easily be "written" or "painted", in whole or in part, by everyone else, without attribution, then go ahead and opt in.
But if an author or artist does not EXPLICITLY OPT IN, you can't use their creative work to train these systems.
All these code/art washing systems, that absorb and mix and regurgitate the hard work of creative people must be strictly opt in.
If you don't care about this, it's naivete, or a lack of foresight, or apathy as these companies pillage the commons. Not something to be proud of.
Microsoft and OpenAI (and others) are robbing us and you should care.
except we are in 2022, the artist died 22 days ago https://news.ycombinator.com/item?id=33105773
and https://twitter.com/bg_5you/status/1578146498768175105 has "credit plz"
so as long as in 2120 the tech is trained on public domain artists and code, it will be ok.
But if it is trained copyrighted sources, AND requires its own attributions without actually requiring the copyrighted source attributions and/or its licenses, then you are right, we won't get it. The same way we have SSPL and other licenses in addition to public domain code.
Pretty much everyone can down and use the publicly available models. And for those without the hardware, there are cheaply available online services you can use as well.
You can use it as much as everyone else. Problem solved.
Like, I can deduce that your argument comes from a place of "this work is mine and mine alone"... but is it? We don't have copyright on mathematics for example, if you come up with an equation or a solution to a tricky problem I can make use of it without the need of attribution. And in many ways, attribution can work similarly in AI generated art; it is common for people seeking a distinctive artists style to write prompts that include said artist and the AI, in turn, will willfully comply.
But what's the real core of your argument, why is it that you think information that is shared in this way is bad?
I can actually empathize with artists that think this technology might threaten their livelihood but this particular position I also view as overly-catastrophic. But you don't really seem to be concerned with that and instead you're making an argument that comes solely from individuality? But to me, these AI models are the maximum expression of something that as societies we have been doing for a very long time: to take the work that's available and use it to create something new.
Now, I completely agree with the licensing issue when it comes to github's copilot. It used public code to train, and therefor any work that is produced through it should be public as well. It's a different issue.
But your argument that "your work becomes my work" is bad is not something I fully understand. Why is it bad? Why is it poison? Shouldn't we maybe be questioning the system that envelops us if such a thing as sharing is considered poison?
Artwork can be very much licensed the same way that code can. If I train my AI on Creative Commons Share Alike or None Commercial licenses then why should that be any different?
And if we honour one sort of licensing, the copyleft side, why should we treat the other side, the copyright differently?
Either AI is allowed to ignore copyright and licenses for everything or it's not, copy or painting, it makes no difference.
This is flamebait, lacks nuance, tone deaf (who still says hippies?), etc. Please try to practice empathy online.
I, as an consumer, don't care if anime/manga/textures/etc... are created via hand or AI.
I just want a good story or good entertainment.
Thus, my vote with money will go to whoever creates the best story or entertainment.
I don't care about you needing 20 hours to create a handpainted manga, I want my entertainment: and if someone who cannot draw uses AI to create a 10x better story than someone who can paint but is shit at stories...well..though luck, I don't care.
Probably true to some extent, but that's also why copyright laws are put in place, so there could definitely be laws made that could require licensing if you use it to generate an AI and then use it for commercial purposes.
The time-window to have the debate of if you should have such laws or not is now.
If someone sets up an AI anime shop in Mozambique, who is going to stop them?
But theoretically, countries can ban your product if it doesn't comply with local copyright laws. That can mean that say any video game, movie or tv show made with AI generated models that break copyright laws in say US could be banned from being sold in the US.
You can even go a step further, and sanction the company, blocking all payment processors that are looking to process payment in the US from disbursing money to that company.
So in practice I think you're right, and I'd advise that artists, developers and everyone else braces themselves to it and find ways to stay marketable and relevant even in the age of AI, but in theory it's just a political thing, and if there was real will to do something about it, it could be done.
That's not how it works. AI is copying/stealing from people who actually create. "Laundered legal liability" as someone here so eloquently put it. When co-pilot reproduces GPL code verbatim, it's not AI, it's just convoluted copyright infringement. I have no doubts this is happening with anime too. The so called AI is just a thief. It's not creating new works, it's copying parts of works and making a composite.
When I write, I am doing nothing but laundering sentences and grammar rules from my English textbooks as well as sampling from the millions of sentences that I have read in my lifetime to compose into this comment as a response to your prompt. When I code, I heavily reference existing implementations and often only make minor stylistic changes, copying algorithms and data structures wholesale from memory. When I cook, I'm literally copying the contents of the cookbooks or cooking channels with a few minor substitutions at best.
Unfortunately, I am not brilliant enough to come up with my own algorithms, recipes, or literary flourishes. Few people are truly "creators", to be honest, and even creators build off or derive from existing work.
Most people don't care about what is artistic. It's the experience that matters.
It's a huge loss IMHO. It's much more substantial than copyright issues and jobs. The artists positions in the article articulate this, for some reason quotes like: “Artists are not just a ‘style.’ They’re not a product. They’re a breathing, experiencing person,” are being read as artists being scared for their livelihood. Why not read it as it's written? As artists fearing for their humanhood?
But I'm pretty sure the reaction of most "geeks" was positive. Geeks generally love new tech, especially when it can improve their workflow.
A very valid negative criticism is that a huge company with a bad track record having ownership of a for-profit tool created by processing the work of millions of people is not good.
However that criticism shouldn't be conflated with criticizing the existence of the tool itself.