If a good artist imitates and a great artist steals, wouldn’t a computer program designed to steal be the greatest artist ever?

Painting in VR, playing video games in AR: the future of art-consumption looks incredibly exciting. But these examples are all rooted in humanity: created for and by humans, a closed loop of foibles, brilliance and stupidity. MONTAG's Sean Fleming asks: what if we left the humans out of the creative process all together?

Talking to yourself

You may have heard that Facebook had to abandon an experiment recently when two artificial intelligence chatbots started communicating with each other in a language they’d created themselves.

The exchange was an attempt to negotiate the trade of items such as balls, books, and hats. The chatbots were successful at trading with one another albeit at the expense of our understanding. The bots had created their own optimised version of English that they understood – but we humans didn’t. Great! Ixnay on the shutting us out of your comms please, Skynet!

If you haven’t seen the exchange between chatbots here it is in all its baffling glory/weirdness/glorious weird:

To me, to you, to me, to you, ad infinitum (Screenshot: Facebook)

I mean, I wasn’t expecting Arrival-tier science fiction nonsense or some binary wonder, but - you know - come on. Half malfunctioning Chuckle Brother-bot, half ball fondler (half pig).

As usual, clickbait content had given me unrealistic beauty standards: every publication talking about the bot made it sound like some mad scientists pulling the plug on a dangerously cogent AI poised to destroy everything you care about.

Actually it just wasn’t doing what it was supposed to be doing, so they turned it off. Thanks, Millennial Media. Doin’ it for tha clicks as per usual.


Once I got past these sensationalist headline-driven expectations, I appreciated the fact whatever computational shorthand the AI had used to conduct negotiations, it had created something new from another input.

Think about that for a second.

It was given access to English but it thought Fuck that! I’ve got something better! You could be forgiven for thinking that inter-artificial communication has an air of humanity surrounding it.

It got me thinking: if an AI was perfectly capable of creating something new from human input – with the result requiring some element of interpretation and so was to some extent a creative use of the source material – then AI could quite legitimately create what could be deemed art.

I'm not a caveman, I'm a cave, man

For me there was a striking similarity to our attempts at understanding parietal art.


Cave paintings, AKA Parietal Art, you cultureless oafs. (Picture by Abdullah Geelah - English Wikipedia, CC BY-SA 3.0)

Cave paintings predate written language and it’s widely theorised that they were the way neolithic humans communicated with one another. Of course, there’s nobody left to ask whether that’s really true or not.

But consider how we modern humans try to understand the lives, fears and day-to-day goings-on of a people so far removed from ourselves: could we view early AI to AI comms in the same way?

We can only interpret, and in our attempts to understand, we mythologise Cave Painting. We elevate it to art. Maybe we can do the same for the early communication between chatbot and chatbot.

Because – and stop me if this is getting too cerebral (see also: pretentious) – there is an innate value in language and communication even when we don’t fully, or barely, understand. Take that Damien Hirst! Your rotting cow is bullshit.

Of course, once you start bunching this bot-on-bot creative phenomena with human art and crediting it with possessing human qualities things get very interesting indeed. (Protip: it’s still better than Damien Hirst).



Technology and art have always had a tumultuous – and incestuous - relationship.

Before photography, painters and sculptors strived to create what we might now call photorealistic forms in their own medium. And then mechanically perfect realism – as brought to you by the camera – put representational painting on what we could refer to delicately as indefinite hiatus.

Painting didn’t just stop though, and photography’s influence on painting gave birth to new forms.

MONTAG has discussed before how AI is coming for your cushty job in creative media. Well, I hope you’re sitting comfortably in your Herman Miller office chair – and polish off that flat white before you spit it all over your Macbook – because AI is coming for your cool mate’s gallery métier too.

Soon, it won’t only be us spending Thursday evenings at gallery openings, drinking shit wine and murmuring how we’re only here because our flatmate is dating one of the light-installation artists. The machines will be processing backhanded compliments from poseurs pretending to understand what’s on display too.

Like acid, on acid... on acid


Most people’s first experience with algorithmically-etruded art was 2015’s Google DeepDream bonanza.

For the uninitiated: DeepDream was Google’s super trippy neural network that everyone was playing with a few years back. People were drawn to it because it was pretty cool tech, yeah; but I, like many other curious pranksters, pumped images into the bot in the hope that the results would help us relive that time we did acid at Glastonbury, one demented pixel at a time.

Generally, I think people were fascinated by the idea of a computer creating art. For me, one of the most beautiful traits of DeepDream’s work was that, by putting in an image, or music, or film through the bot, whatever it spat out was its attempt to understand the artistic input.

It looked for patterns and shapes the same way humans do. In all the white noise it latched on to stuff it vaguely understood at a very base level, which much like myself is mostly just dogs’ faces and eyeballs.

Crank up the Aphex Twin: we’re going for a ride!

Perhaps as humans, our best route to understanding our arty bot buddies is to collaborate with them on an project - you know, you like you did back in art school with that cute weird foreign exchange student.

And just like back then, hopefully the act of teaming up will lead to a lot more exciting stuff happening, at a much more primal level.

Perfect Harmony

Studies are already suggesting that the future of work will be a sympathetic balance: AI and humans working side-by-side rather than one pushing the other into unplanned obsolescence. Perhaps art will head in the same direction.

One artist using AI to understand his own work and then produce something new off the back of this new perspective is Roman Lipski.

The concept was simple: have a machine learn Lipski’s painting and in doing so give the machine what can only be described as an ‘artistic intelligence’.

Having learnt from Lipski’s paintings, AIR (Artificial Intelligent Roman) created new work using his style and motifs for Lipski to use these new paintings as inspiration for new work. Yeah, exactly.

LILO. Lipski In Lipski Out.

As the first example of an artificial muse it’s a fascinating project, and on a technical level it's a brilliant example of using data and a non-human perspective to better yourself as an artist.

Much like how photography pushed painting into new and exciting realms, the advent of AI-derived artistic produce should make artists - and art - better.

Lipski’s AIR also raises a few interesting questions, the most basic being: if Roman then signs one of these artificially-created paintings, does it become his work, as if he’d created it by his own hand? Why not? How far removed is this from Andy Warhol’s soup can screenprints?

If a good artist imitates and a great artist steals, wouldn’t a computer program designed to steal be the greatest artist ever?


When an artist's work can be fed through a machine that understands it and produces new work to rival the quality of the original, can we program artistic sensibility into it too? And can it then produce its “own” work?

Humans create art to start dialogues, to provoke, to speak. Can something that doesn’t have human-like emotional intelligence ever produce something really worthy of the title art?

If you saw a piece of work in a gallery and the artist that had created it was none other than the RothkoBot 3000, would the cold, silicon mind behind it devalue the work, in your eyes? The work produced an emotional response: does the fleshiness of the creator even matter?

And just think of the investment opportunities: maybe a robot’s art would increase in value every time it gets a firmware update. What if the artwork itself was updated over-the-air every now and then?

AI-produced art triggers so many questions, although not all of them rhetorical, so feel free to whip these out at dinner parties to one-up whoever’s hosting.

Our motto: Artpocolypse Now!

Maybe we should all accept our fate, and get ready to embrace a future where art is better, cheaper, and everywhere.

After all, AI is going to be better than us at almost everything, and if it’s better than us at the most human expression of all, then hook me the fuck up. Walter Benjamin, eat your heart out.