Christie’s sells its first AI portrait for $432k
theverge.com
theverge.com
We changed the above url from https://www.bbc.co.uk/news/technology-45980863 to one that has a bit more info.
GitHub link: https://github.com/robbiebarrat/art-dcgan The README was last edited 6 months ago and contains the bolded text: "When using any outputs of the models, credit me. Don't sell the outputs of the pre-trained models, modified or not. If you have any questions email me before doing anything."
https://www.geek.com/games/box2d-creator-asks-rovio-for-angr...
The whole discussion is meant to elicit "a pass". Dishonest.
So Hugo asks for help with setup because he can't get it to work, and Robbie (original author) replies "working on a rewrite of the codebase, should be soon" while Hugo keeps pushing for updates. Turns out it's trickier than expected so Robbie says "sorry, might take longer". Hugo's response?
> "Ok, really disappointed. Could you at least provide the code for the 128x128 Gan in Keras ?"
He's literally sitting there for months pushing for poor Robbie to do the free work for him so that he can try to profit from it.
Maybe they really did make a lot of changes to the model. But from the conversation in that issue thread and the striking similarities between the painting and the other ones that Robbie had made over a year ago, it doesn't feel likely to me that that's the case.
Hopefully some justice will come out of this situation.
As an aside, according to his bio Robbie's only 19 (and already doing this level of research!). Sad to see younger researchers being taken advantage of like this.
https://github.com/robbiebarrat/art-DCGAN/issues/3#event-140...
Also the DCGAN architecture is from Chintala et al, and that itself is based on work by Goodfellow et al. Where does this end? Does Robbie Barrat 'own' the DCGAN NN architecture now?
And enough people don't understand this for it to be a problem.
Edit: Removed bad phrasing.
It's the main reason there is a widely used public web rather than it, if it succeeded at all, being another academic niche tool alongside gopher.
The French kids owe the Stanford kid like half to all of it.
Edit: Actually, there was a change to the license surrounding pre-trained models and sales back in June, so it's even baked into an obvious contract. A lot of the licensing conversation here is somewhat mooted by this.
https://github.com/robbiebarrat/art-DCGAN/blob/master/LICENS...
It sounds comical that suddenly something becomes valuable because you hang it in a gold frame, make minor modifications to it and ascribe it some symbolism. But that is exactly how contemporary artworks are done.
If you go to exhibits and auctions, it is not uncommon to see highly valued work that requires little sophistication to produce. The days you would woo the art world by technical mastery alone are (mostly) over. This story fits in the greater trend and there might be a bit of a culture clash.
I'm inclined to think that the general public underestimate contemporary art and the art world overestimate its superiority over the more conventional works. Whether it is for better or worse, I will leave you judge of that.
I really like your framing of this and agree wholeheartedly. I think it really comes from contemporary art’s focus on abstraction and meaning within context. The combination of focus on abstraction and meaning while requiring the viewer to understand the context in which a work was made makes it very inaccessible to all but those who have the time and knowledge to develop that understanding from. This makes it very inaccessible to general audiences. Alternatively, so often the depth of the meaning/context and the effectiveness of the abstraction are quite underwhelming in reality but overly celebrated by the art world.
It's got to the point of satire:
If you spent two minutes rolling some white paint on a white canvas because your "abstract message" was about the void of space, it's not profound, you're being an ass.
Conversely if you spent at least five hours stippling black spots on a white canvas in the image of some obscure shape, because you had a concept in your head, and it took a considerable amount of effort, THAT is a profound piece of work.
The art world isn't full of bohemians with time and knowledge who've developed understanding of abstractness. It's full of pretentious, bored, rich people. Huge difference.
Got it. work >= 5hrs == TRUE REAL PROFOUND ART, work < 5hrs == UTTER GARBAGE PRETENTIOUS BULLSHIT.
Art is purely expression. It doesn't matter how long it takes you to express what you're expressing. Pollock may have just splattered painting on a canvas. But it was still a true expression. John Cage made a composition where the orchestra doesn't play any notes. But it's still an expression.
The fact that rich people are willing to pay for some bland expression does not confer anything but monetary value to it. It doesn't become less garbage.
Sure, but that in itself doesn't make it any less art. Saying "I find this art boring and trite and consider it garbage" is fine. Saying "I find this boring and trite and thus it's garbage and not art" is in my opinion not a valid argument.
Cage made a recording of an audience waiting for music. The audience was the work, not the orchestra.
That said, effort isn't generally what makes or breaks art, either in the gallery / collector world, nor among artists. Very, very few stippled paintings that took more than 5 hours of work are considered profound in any way. It's been done already.
Maybe you meant to imply skill, but skill isn't that important to art history either. Somewhat important, perhaps, but not generally in the top 3. Concept and narrative and novelty and luck all outweigh effort and skill, even among unpretentious skilled poor artists.
> The art world isn't full of bohemians... it's full of pretentious, bored, rich people.
I'd be careful to qualify that statement. There are lots more poor artists than rich people. When you say "the art world", if you're talking about Christie's and Sotheby's, then yes there are rich people. My personal "art world" is full of artists, and I don't pay attention to the stuff the auction houses are auctioning.
That's something one might conclude about the "art world" if their only impression of it is from hearing about high profile auctions of weird paintings sold for hundreds of thousands of dollars, which fits nicely in the general "rich people do the darnedest things" category of trivial news. Sure, someone with the right name can probably spit in a napkin and buy a car for the money, but in my experience this is a minority overwhelmed by people that can't even support themselves with their art, much less justify the cost of enjoying expensive works of art anywhere but in a gallery.
I know many artists with vastly different approaches to expression, none of which are rich, most of which also have to work day jobs to make ends meet. I don't think any of them would regard a white-painted canvas as particularly interesting.
Most halfway decent art galleries and museums are pretty good at providing very readable summaries of the exhibits they're putting on. Spending just 10-15 minutes reading the introductory text the museum provides is often all the background knowledge you need to gain a much greater, if not appreciation, then at least understanding, of what you are looking at.
Ah, love it.
People often mistake art for a form of labor production. They expect that labor has to have been performed to create value. But that all changed in the early 20th century.
I really feel bad for Dr. Beef, this must really hurt. But according to the twitter thread ( https://twitter.com/DrBeef_/status/1055285640420483073 ) they didn't use the pre-trained models. In any case pre-trained models are labor, not end product. They simply trained them from scratch. Possibly on the same input set.
Even if his license was more strict, they still could have just rewritten the same code and it would have escaped the licensing. Should the original GAN paper ( https://arxiv.org/abs/1406.2661 ? ) be given artistic credit here ? What is authorship in this context ? There is a whole chain of people furthering things and doing labor and making decisions.
Yet nobody has even mentioned the original artists whose works were used to train the model. Would it be different if they were living artists ?
In what way is a GAN essentially different than photoshop copy and paste of elements ?
I may be misinterpreting your intent, but this seems disingenuous to me. I can't go look at the Linux kernel, then 'rewrite' [the same code] with my fingers and keyboard to escape copyright ownership of Linux kernel contributors.
I can rewrite code that implements a described process. Or code that implements a described interface. If this was your intent, I'll take the blame for misunderstanding. But I caution the use of phrases like "the same code" when you actually mean new code that implements the same process.
> narrator/reviewer considers Menard's fragmentary Quixote (which is line-for-line identical to the original) to be much richer in allusion than Cervantes's "original" work because Menard's must be considered in light of world events since 1602.
To generate the image, the algorithm compared its own work to those in the data set until it could not tell them apart.
Hopefully the person who wrote the code, freely shared it and even helped them to get it running will see some of that ridiculous windfall:
Alternatively, the Obvious collective will be now and forevermore tagged as non-contributing leeches. Which will likely not tarnish their enjoyment of their $144k each before auction fees and taxes, but might discourage future AI/ML developers from releasing under such a permissive license in the future.
Exactly. It doesn't sound like they are afraid of this but they really should be. Nobody will want to buy the next artwork of a collective with that reputation.
I don't say this because its my personal view -- every open source project I've used I try to contribute to in one way or another, and actively ensure that the author's receive fair credit even when their license does not require as such. But my view is based on reality. There are many people that are not like this. You give them something and they will simply take it. The actual creator or provider of what was given will be out of their mind before nightfall. Or in a work environment you may see people work themselves to death trying to get ahead only to find out that the person one notch ahead of them on the hierarchy had been happily taking credit for everything they were doing, and reaping all the accolades for such.
This is something that I think people should consider much more carefully before deciding to make their code freely available. You may indeed make society a better place, but there's a very good chance that you will receive no credit, let alone reward, for your role in it. If you consider that acceptable then take it a step further to consider that somebody else will, not infrequently, end up taking credit and reward for your work. If people were naturally driven to always want to reciprocate with reasonable value, the history of our species would be rather different.
In this case though, it looks like Obvious did set out to attribute the original GAN author and Christie's have omitted that.
It’s not legally required to credit the author of the original code, but it’s a jerk move not to do so, especially if the work is primarily not your own.
All it does is discourage developers from releasing things open source, which is very bad in the long run.
What on earth are you on about!? The author has every possible opportunity to set out whatever he wants in the license. If he wants anyone who uses it to name their firstborn after him he can put that in there. If he doesn’t then there is absolutely no issue or nothing jerk-like of anyone who follows to the letter he license the author decided to use.
It’s up to the author to specify what he wants, and as long as users follow the license that’s their right.
If you want a license that only allows non commercial usage to be unattributed but requires attribution for commercial usage... then use such a license. Don’t just put it up under a permissive license and rely on calling people jerks when they literally do what you specified and gave them license to do.
> all it does is discourage developers from releasing things open source...
Great! Developers who don’t want others to use their software under an open license shouldn’t be releasing it under an open license in the first place! You have all the control as a creator, don’t give it away and then complain that you gave it away like it’s someone else’s fault for using the license you gave them.
This is the same trap I see 40-somethings falling into time after time, acting like AGPLv3 is evil while complaining how their (non-copyleft, open source) projects are widely used, extended and not contributed back to by those that use them. If your building a totally new platform in an area with no existing tools, strong copyleft is not a bad choice.
I hear an increasing amount of talk as if ethics is or can be entirely contained within law, as if behaving legally is all that can or should be asked of us as humans. I don't know if it's the pervasive influence of lawyers and corporate 'culture' or what. In that "greed is good" etc mindset, 'being a jerk' isn't a bad thing, it's merely not being a sap, using your freedoms, using accepted business principles etc. Your whole comment smells of that total absence of ethics, beyond obeying the law, i.e. no ethics. Sorry if I misunderstood. Maybe your "What on earth are you on about!?" etc was exaggerated for rhetorical effect, and not what it seemed - having no clue why someone might call it 'a jerk move', or feel it was unfair/rude/graceless/exploitative etc.
There’s nothing unethical about respecting people’s freely chosen and publicly expressed wishes.
Is it, though? In terms of keeping a health job market, I'm somewhat skeptical that offering things license to allow even commercial usages for free is necessarily a long-term good for me.
Why is it being a jerk to engage in a behavior that someone freely gave their blessing to?
The problem with this is that it disincentivizes the sharing mentality. Nobody wants to be the guy who gets taken advantage of, yet everyone benefits from free sharing of knowledge. And then once in a while you have dudes who swoop in and make everyone look stupid and make them re-think whether they should ever share their findings with others again.
Obvious could make this right if they wanted to. Yet they are standing their ground on Twitter, it seems.
This loses the important nuance that them swooping in and slapping a price tag on their own use of the implementation doesn't deprive anyone of access to the original on the same terms they were able to access it.
No one has lost anything in this scenario.
>The problem with this is that it disincentivizes the sharing mentality.
Personally, I feel arguing over how people make use of freely shared knowledge does more harm a culture of sharing than usage of said knowledge, even when that usage pushes right up to the bounds of what is allowed under whichever license.
This is a really horrible life lesson that licenses matter and if you create something it’s important to think through yourself how to make money off it and do it yourself (and think hard if you care about that because caring retroactively is too late), before you set it free because if you don’t do it then someone else will and honestly you only have yourself to blame because the reality is you didn’t want to put the small bit of extra effort required to make money off the thing
I agree with you, except for the words "small bit". As an inventor, with a few issued and pending patents under my belt, I can assure you that bringing an invention (software or hardware) to market is no small feat.
For me, concieving of crafty algorithms or a new device is the easy part. Even writing a patent, whild time consuming, is straight forward. However, getting people to buy said inventions and establishing an efficient "supply chain", for me, is the excrutiatingly hard part. For some, this part comes naturally, and it's the act of invention that is tough. For a lucky select few, they are talented at both. Still, the requisite investment of time and money is generally not small.
Try to sell identical thing on eBay for $10 and watch the world making fun of you at zero bids.
I don't have much sympathy for the sellers profiting off open source code without apparently any intention of contributing back even a little, but it's clear that the 400k valuation is not a direct outcome of the code quality - after all you can make litteraly millions of painting from the same "author" for bucks each. The 400k come from the marketing skills of the sellers (or, if you're less generous, from the fact that they're probably very well connected to the right people in the art market).
The real feat for me isn’t in using this open source model to generate this painting, it seems to be convincing anyone that this was worth $432k
It's just what a segment of the art world does - mess with the definition of the art world. Look at Banksy's million dollar shredder.
I wonder if this is a marketing stunt, or an effort to increase the visibility and legitimacy of AI generated art.
You heard it here first!
For $400k you could buy a dozen works by some really influential artists. But there's a big pool of money willing to buy work made by some 25 year olds that is totally disconnected from any broader cultural apparatus related to art. I'm fascinated, and wonder how the art world will respond. Clearly there is money to be made.
If you wanted, you could print your own!
We know math and science can be found within art: measures and half measures in music, the golden ratio in painting, etc. Optimizing for these mathematical properties seems a bit like cheating for lack of a better word. There are many musicians and artists without classical training that encounter these attributes on their own, and manipulate them in different ways. I think part of the reason why the arts are such a great platform for human creativity is because of this.
Does they though? I bet in 10 years, if you walked into an art gallery, you'd still be drawn to certain pieces and not others regardless of wether they were created by human or AI.
As AI improves it will get ever closer to that "hand-made" look to the point we probably won't be able to tell, especially with something with as much intrinsic variety as creativity. In fact, if AI is trying to copy a "human style", we might be able to define how narrow or broad our range of creativity really is.
At it's current state I have doubts about AI's ability to create new movements in art as well, a lot of the AI floating around there is very good at "coloring within the lines". If I watch a video of NFL highlights on YouTube I will be recommended more NFL highlights, where is the discovery? A fundamental challenge for AI art is can it create new genres. Could an AI create Jazz? How about Hip-Hop which has some Jazz elements. Could an AI create cubism? Guess we'll have to see.
I am skeptical how much art by AIs will actually feature in museums and exhibits. High-caliber art is very reliant on its emotional context and the message that the artist is trying to convey with the piece. However, I could see AI art featuring when an artist invents a new algorithm or new AI-based technique. Again, going back to the context of the artist and the piece itself.
That debate goes way, way back. In 1966, a researcher at Bell Labs used a computer to generate an image in the style of Mondrian. The images were presented to 100 people; not only were test subjects unable to identify the computer-generated image, but the majority liked the computer-generated image better. The computer-generated art is now in the Los Angeles County Museum of Art's collection.
http://thekidsroad.com/assets/images/blog/blog-content/mushu...
Brian Eno quote about a related field: music "The great benefit of computer sequencers is that they remove the issue of skill, and replace it with the issue of judgement"
The short explanation is: Train and freeze the most basic layer of the model progressively to "let the network" understand higher resolution concepts one piece at the time, and avoid mode-collapse.
The network architecture illustrating this a bit better is shown on page 3 of the paper [1]
That is news, the fact that it is in the news automatically adds to the cache around the piece. That makes for a solid art investment.
I had my portrait painted by a robot with a camera and some variety of edge-finding DSP at the London Science Museum in about 1990 - and they touted that as “AI art”.
Also - what about Conway’s game of life? Is that not art?