Live data from Hacker News

Using Artificial Intelligence to Augment Human Intelligence

distill.pub

11–20 of 33 posts

Re: Using Artificial Intelligence to Augment Human Intelligence

#11
post #9

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

When people say "but will AI ever be able to produce art?" I think works like yours is a good thing to point to. In the end, whatever art will be produced by machines is somehow going to be initiated by a human hand. Your work makes it trivially clear that the answer to the question is (1) yes, and in fact they already are, but (2) there is still going to be a person who will need to use his or her personal taste to…

Already starting in music

https://qz.com/790523/daddys-car-the-first-song-ever-written...

Re: Using Artificial Intelligence to Augment Human Intelligence

#13
post #9

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

When people say "but will AI ever be able to produce art?" I think works like yours is a good thing to point to. In the end, whatever art will be produced by machines is somehow going to be initiated by a human hand. Your work makes it trivially clear that the answer to the question is (1) yes, and in fact they already are, but (2) there is still going to be a person who will need to use his or her personal taste to…

"but will AI ever be able to produce art?"

I think the modern definition of art is that art is anything that the observer perceives as art.

"and in fact they already are, but (2) there is still going to be a person who will need to use his or her personal taste to filter out the bad from the good, which might be what is going to replace the old way of creating things from scratch, e.g. with pen and paper."

No. Currently AI are synthesizing extrapolations from existing data. They can paint like van Gogh only because they have sufficient samples of the van Gogh style to learn the style transfer. Currently you still need a human to invent the style, at least.

On the other hand, once a style is invented, it's a matter of taste I suppose where the border lie. Ancient painter masters did hardly paint everything by themselves. The most prolific were more like Walt Disney or Damien Hirst who employed and employ legions of workers to implement their vision, and the artist merely signs off sufficiently high quality work.

So, I suppose you are wrong and correct at the same time.

Re: Using Artificial Intelligence to Augment Human Intelligence

#14
post #9

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

When people say "but will AI ever be able to produce art?" I think works like yours is a good thing to point to. In the end, whatever art will be produced by machines is somehow going to be initiated by a human hand. Your work makes it trivially clear that the answer to the question is (1) yes, and in fact they already are, but (2) there is still going to be a person who will need to use his or her personal taste to…

The same people said Pink Floyd (and other early electronic music artists) aren't musicians because they use synthesizers. It's fundamentally misunderstanding what AI is (and what art is too, for that matter).

Re: Using Artificial Intelligence to Augment Human Intelligence

#15

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

Please take the following as a form of intellectual sparring, and spoken loud hypotheses about art theory.

"I processed pictures using the Artistic Style Transfer algorithm and then painted them."

Isn't that just effectively painting by numbers? Autotune for visual arts... and it's not interesting once one learns to percieve the effect.

The problem is that one takes a less interesting input domain, and runs it through a filter that makes it more aesthetically interesting. Once one learns to percieve the common transfer function my brain filters it back into the less interesting input domain. Music and visual arts as well.

When digital images became common there was a huge influx of people who ran their images through photoshop filters and called their output "art".

What one likes, one likes, I have no qualm about that.

For me personally, though, that is just tastless, unimaginative and boring. Maybe it's because once people use well known domain transfer tools, my brain learns to discern those domain transfers and robs the output of it's pleasant features induced by the domain transfer.

To provide actual value, the process should make the content somehow more "alive" or "interesting" than just by applying a "trivial" domain transfer, like a well known neural network domain transfer or photoshop filter.

Maybe a person should invent ones own domain transfer system, thus not make it commoditized?

Many ancient master held their work strictly in secret before it was ready. Maybe, they realized that if the process was known people would recognize it simply as consequent application of various domain transfer functions and the magic would be lost.

Re: Using Artificial Intelligence to Augment Human Intelligence

#16
post #15

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

Please take the following as a form of intellectual sparring, and spoken loud hypotheses about art theory. "I processed pictures using the Artistic Style Transfer algorithm and then painted them." Isn't that just effectively painting by numbers? Autotune for visual arts... and it's not interesting once one learns to percieve the effect. The problem is that one takes a less interesting input domain, and runs it throug…

Your last paragraph is right. His only mistake was to publish his creative method. If you just stumbled upon his art in a gallery you would have no idea how those paintings came to be.

Re: Using Artificial Intelligence to Augment Human Intelligence

#17
post #15

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

Please take the following as a form of intellectual sparring, and spoken loud hypotheses about art theory. "I processed pictures using the Artistic Style Transfer algorithm and then painted them." Isn't that just effectively painting by numbers? Autotune for visual arts... and it's not interesting once one learns to percieve the effect. The problem is that one takes a less interesting input domain, and runs it throug…

Style Transfer is just as much Autotune for the visual arts as Spray Paint is. The process doesn't matter, only the end result and the impact that result has matters.

The means to achieving that impact is an art unto itself divorced from any resultant work; a meta-art, if you will, which is deserving of its own consideration and appreciation.

Re: Using Artificial Intelligence to Augment Human Intelligence

#18
post #3

The quality of this journal is mind boggling.

Programmed frontend demos here and in places like news sites can be used to great effect. I how one could increase accessibility of these sorts of widgets to non-programmers? Deeplearning is a bit out of scope for that, but interesting graphics have a bunch of other uses.

Re: Using Artificial Intelligence to Augment Human Intelligence

#19
post #15

My work is very relevant here. I'm a graffiti artist and a programmer. I started using machine learning techniques to augment street art with various nature landscapes (drone images, rolling hills, etc). By using simple techniques, such as Style Transfer, I have been able to create some interesting images that are aesthetically pleasing, but unusual. I took the digital results and painted the contents on large canvas…

Please take the following as a form of intellectual sparring, and spoken loud hypotheses about art theory. "I processed pictures using the Artistic Style Transfer algorithm and then painted them." Isn't that just effectively painting by numbers? Autotune for visual arts... and it's not interesting once one learns to percieve the effect. The problem is that one takes a less interesting input domain, and runs it throug…

Honest intellectual sparring over art comes in the form of art. Saying "I could have made that" is false in the context of art. If a person could have they would have. If they didn't, they can't. The proof is in the work.

Re: Using Artificial Intelligence to Augment Human Intelligence

#20
post #3

The quality of this journal is mind boggling.

Brett Victor has done some lovely stuff allowing people to interact whilst telling a story [0]. Doesn't work so well in mobile, but the Google page suffers in this aspect too.

[0]: http://worrydream.com/LadderOfAbstraction/

Post reply on HN