Can Computers Create Art?

Can Computers Create Art?

I’ve been watching this nice talk from Aaron Hertzmann.
While I’m not sure I agree with all his points, I appreciate the discussion:


“Art is about intent” -> guy comments “it’s almost a religious belief”.

No dude, it’s called philosophy. Perhaps it’s not on the Silicon Valley 101 curriculum, but shit lying outside of the range of menial pragmatical technics doesn’t automatically qualify for cult or belief.

Always amazed by the idiocy of brilliant people once you poke them out of their everyday obsession. At least fishes out of water have the decency to shut up.

Computers create graphics. Until perhaps they grow a conscience.


Here is more fuel for your annoyance. :smiling_imp: We now have bots writing better articles than some humans. I am sure they are already writing art reviews about AI-generated pieces. :robot:


“Better” is contextual to a goal. There is no “better” if you don’t state the goal. My guess is articles are supposed to be factual and informative. Art lives on another plan.

1 Like

Exactly!! I love Bob Dylan’s music and follow a couple “fan” groups on Facebook. Every day, multiple people post topics either broadcasting or seeking opinions about which album/song/era is best. Whenever I have the energy to post a reply my answer is always, “Best at doing what?”. Nobody ever responds :laughing:

The ACM tends to say no they can’t, or at least they publish articles saying so:

As we get people using tools to generate other tools to generate or create art it’s just a question of who’s more responsible. Clearly the tool is important but only so far. I have nice cameras but am a terrible photographer, also bad a coding despite a good keyboard. However as the tool ends up doing more of the “driving” I think there’s probably question around how much the tool creator contributes versus the relatively hairless ape pressing the button. I’m thinking things like computational photography in phones. The person took the photo and someone wrote the code for things like on the fly tone mapping and the fake bokeh. So who gets more credit? Your vision is influenced (especially since almost everyone is using EVFs now) but the image the tool presents you so how much credit that person in R&D gets is an interesting problem as these tools advance IMO.

1 Like

This was meant to be quip on bad human writing. I recall the moment in middle school when I no longer found newspaper and journal subscriptions worthwhile because of their poorly conceived articles. The advent of writing aids and social media certainly made it much worse over the years. Coming across good paid writing is uncommon and a pleasant surprise nowadays. This is especially true in the art world where people pretend to be sophisticated with exhausting language and misused vocabulary.

PS There have been many accounts of researchers who have proven this by submitting fake articles and papers to news agencies, academic journals and reviews. Poorly written and full of BS but apparently enough to bypass the review and editing process with flying colours.

If a computer “makes the art” a human still has to select, curate, display and view the art. Was in the guys in the factory making Fountain or Marcel Duchamp who was the artist? I use that example because it’s famous and explicit. So much art is about picking up, selecting and arranging existing things (photography?)

The question of intent is slippery because things not made as art can become art at a later stage. But usually as part of a collective or individual select and arrange.

The video (didn’t watch the whole thing) points out that only a social agent can create art. Without excessive thought I think this is correct. Intent is to vague and fleeting requirement when considering a fair bit of existing art.

A tool is a tool. Something that does the driving is an AI, not a tool anymore. A tool responds to the person holding it.

So who gets more credit? Your vision is influenced (especially since almost everyone is using EVFs now) but the image the tool presents you so how much credit that person in R&D gets is an interesting problem as these tools advance IMO.

That becomes an Oscar joke : “I like to thank my mother for giving birth to me, my first grade teacher for teaching me, Jo the taxi driver for getting me on time on my first day, etc., etc., etc.”. The credit goes to the person who used tools that could have been used by anybody else to get a result nobody else got. Fuck R&D, fuck technology, they are all means used to provide options to creative people. What matters is how they are used. We don’t credit the brush or the canvas maker for the Joconda…

I think the point is not so much about social agent than about agent with conscience. Also intent is a property of art (which is the nature of the thing) that was brought recently in art history, especially because technology forced us to reconsider art (see Walter Benjamin, The Work of Art in the Age of Mechanical Reproduction, Art has been a self-reflexive object for a couple of centuries now, as in “artists know they are doing art and do it with personal ethics”, and you find many movements that enforce a certain idea of what art should be, so you can’t look at past art with a uniform definition in mind.

IMO that’s a gross oversimplification. The days of a tool just relying on inputs from the end user are receding for most people and not just in the arts. The tool looks at the scene, makes a decision based on a pre-programmed set of variables and presents an image to the user who moves the box around and takes the shot. There’s no way the user’s vision isn’t becoming more and more influenced by the tool nowadays. Yeah, it’s always been the case with lens selection and viewfinders presenting a certain “look and feel” but it’s cranking up to 11 now with computational photography. You still need a primate pushing the button and aiming the camera, but it’s increasingly becoming complicated when the tool is altering the image on the fly based on things the artist cannot change. I don’t mean this as any sort of value judgement just an observation. Tool, tool maker and creator is a more interesting relationship that it was 200 years ago or even 50 I think.

Granted I’m a awful photographer and I instantly correct anyone calling me an artist. More and more lately been considering hanging the camera up, but that’s the topic for another chat. Literally anyone could reproduce what I crank out. What’s the saying “there’s nothing unseen in the world?” I can’t speak for everyone but my particular vision isn’t really unique or something that couldn’t be replaced by a few poorly written Perl scripts. I guess what I’m getting at in this is not all photos are art. I’m trying to discern between the baggage associated with the term artist and the more technical nerdy aspect of photographer. I’m more the latter. Really a tangent and not related.

I’m also a giant moron who should probably learn not not argue with smarter people and just listen. Feel free to ask if I’ve eaten any good books lately. :stuck_out_tongue: I’m more of a tradesmen and use a lot of tools. Anything above about a third grade US (so probably more like 1st grade for the civilized world) reading level I’m not going to understand.

Now to work on those Perl scripts …

Meh. The decision made is merely how much light goes in and where the sharp object needs to be, for which the user still has to select the optimization strategy (still vs. moving object, width of the object, anchor either aperture or shutterspeed, or max ISO in case of P mode, etc.). We are still far from a clever box, all it does is optimizing stuff based on constraints set and defined by the user.

The complication is a tree hiding the forest hidden in a black box. You got essentially much more steps behind the user and the same 4 dumb parameters (exposure, shutter speed, ISO, focusing plane), but these are the same 4 dumb parameters since the beginning of photography. And if you shoot RAW, you are not really concerned by all the black magic.

I mean, sure, it has probably impregnated and conditioned consumer’s vision, but the RAW (and especially opensource) path gives you all the required options to break free from that and get complete control to convey your vision. It feels overwhelmingly complex because of all the intermediate layers that have been added between actual control and user, but use a camera in manual mode and RAW, and you are essentially back to the basics.

The problem of many photographers I have met is they are too hooked up on reality. And them being usually engineers/scientists only makes it worse. In a regular curriculum, kids gradually stop using their imagination when they leave primary school. Then, they learn to efficiently conform to patterns and methods, so they gradually lose touch with their silly imaginative side to become boring adults with only down-to-earth and pragmatical gears in the brain. (Side note : I wonder if imagination is not correlated with abstract thinking too, which would explain a lot).

I guess the primary challenge of any wannabe artist that had a previous life is to wake those imaginative skills up and become a kid again. Then you will find plenty of silly ways to use a camera that will be more personal and more fun.

1 Like

What’s art is in the eye of the beerholder!

How about Mandelbrot? (visualized math equations)

How about dogs? :dog:



Asking if computers can create art and associating it to a human creating art is like asking if a human brain can create art.

A brain sits inside a skull that rides a top a whole body but just because that body manipulate some medium and arranges and rearranges hue value contrast shape form symmetry, that result is not universally recognized as art.

The art is communicated through the medium from one consciousness to another consciousness.

Humans are expert at recognizing pattern. Humans are also subject to their own perceptual expectations of reality. It doesn’t take much to suggest that a particular pattern of reality exists and the human brain will simply accept it as a reality.

Take the work of Jackson Pollock for example the way that he manipulated the medium in a seemingly random fashion. For all obvious intents and purposes there should be no way for him to communicate any sense of an idea by dripping paint all around the canvas, and yet when a person views his artwork they cannot help but feel that they are being communicated with on a subconscious level.

So my answer is no, no computers cannot create art they can re-create patterns but they don’t communicate ideas.

Here’s my take on the matter. Computers cannot initiate the creation of art since, current, computers are just dumb computational machines. It takes someone who wants to use a PC to actually guide the PC into producing artwork, thus, a PC is just today’s equivalent of a painter’s brush. A brush cannot on its own make art either. There’s always going to be a human being that inspires any artwork to be produced. Saying all this, Deep Dream Generator does automate much of that effort, but even it only produces art based on a human beings selection of presets. :slight_smile: