Previous: Art is Pretentious* | The Art Assignment | PBS Digital Studios
Next: Art Trip: PST: LA/LA | The Art Assignment | PBS Digital Studios



View count:94,877
Last sync:2024-04-08 06:15
When art is generated by Artificial Intelligence, what or who can we call the artist? We look to art history to consider the long collaboration between humans and technology. To support our channel, visit:

Thanks to our Grandmaster of the Arts Indianapolis Homes Realty, and all of our patrons, especially Patrick Hanna, and Constance Urist.

Generate your own Deep Dream image:

More about Magenta and Google Brain Team initiatives:

Subscribe for new episodes of The Art Assignment every other Thursday!

Follow us elsewhere for the full Art Assignment experience:
Response Tumblr:
and don't forget Reddit!:

 (00:00) to (02:00)

(PBS Digital Studios logo)

The Google Brain team drives many artificial intelligence initiatives, including Magenta, a project that uses machine learning to develop algorithms that can create art and music.  Magenta offers tools artists can use to make their own work, but also gleans data from those artists that it then uses to enhance Magenta's machine learning, in attempt to figure out whether machines might actually create compelling art on their own, and it's the two terms 'compelling' and 'on their own' that I'd like to focus on, because in my book, yes, machines make art.  If you'd like to identify as art this terrifying image I created by feeding a selfie into the deep dream generator, a neural network trained in this case to see dog faces in everything, then good on you, or perhaps you really respond to this award-winning machine-generated sonnet that was judged against poetry created by humans which concludes with the couplet 'Behind the wall silence alone replied, was then even the staircase occupied?' or maybe you really like this music video by Taryn Southern, whose music track and video art were created using artificial intelligence.  If you think it's art, then it's art.  Video over.

But the better questions to my mind are: Can we really consider any art to be purely machine generated without a great and powerful Oz setting it into motion or at least operating somewhere in the background?  And more importantly, what is it that we find compelling about art made by machines, or put another way, where is the art in machine-made art? 

If we look back through the history of humans making things and other people appreciating them, we see a long line of artists embracing technology, starting with very basic things like paleolithic humans blowing pigment through hollow bird bones to create stencil handprints or Greek craftspeople using compasses to trace perfect circles and innovating kiln firing techniques to produce the characteristic red and black colors of their pottery.  

 (02:00) to (04:00)

Artists have used a huge range of tools to mold and manipulate materials, render objects more realistically, and do things that the human body alone just can't.  Even if we're not stretching our definitions of technology or machines, you can think about artists as early as the Renaissance using camera obscuras as drawing aids.  A number of scholars believe Johannes Vermeer likely used a camera obscura to help achieve his incredibly lifelike sense of dimensionality and perspective.  

As soon as photography was invented, artists began to use it in their art-making process or as their art-making process.  For a long time, photography wasn't considered an art in itself, too easily achieved, some argued, by the press of a button.  Automated making was often associated with, or relegated to, the realm of craft. 

The Jaquard Loom, invented in 1804, is a machine controlled by a chain of punch cards, strung together in a sequence, allowing the weaver to program the loom to create complicated patterns and effects.  This innovation is considered to be a critical step toward computing, but did the punch cards make the textiles or did a human?

As the 20th century progressed, we see more and more instances of artists making what we can think of as art machines.  Sure, there was art made about life and the machine or industrial age, but there were also electrically powered kinetic sculptures like this one, made in 1930 by (?~3:23) Laszlo Moholy-Nagy, who believed that art and technology working together had the power to transform society in a good way.

For Swiss artist Jean Tinguely's 1960 work "Homage to New York", he set up a self-destructing mechanism in the sculpture garden of the Museum of Modern Art in New York.  It performed for 27 minutes and afterward, the public were invited to pick up remnants of the machine and take them home.

Jumping forward and skipping loads of fascinating instances of mechanical art, there's Roxy Paine's "Scumak" devices, the first of which was made in 1998, which pour hot polyethylene onto a conveyor belt, where it pools and builds and hardens into globular forms, each unique but each the result of an identical mechanical process, or Paine's "Painter Manufacture Unit" whose computer-controlled nozzle attached to a vat of white paint intermittently sprays a canvas resulting in a series of paintings that look like this.

 (04:00) to (06:00)

In each of these instances, the authorship is relatively clear.  The artist made or programmed the machine that does the thing and they attach their name to it as its generative force.  Even if the results are largely or even completely out of their control, whatever happens is still the result of a human being, the artist, having put it in motion.  They've embraced chance, randomness, and in some cases, geology and gravity, to challenge our notions of originality and still lay claim to the result, but where or when is the tipping point between when humans use AI to make art and when the AI can be credited as the artist?

Is the algorithm making the decisions or the person who designed the algorithm?  We tend to think of AI as separate from us and in the sense that we can't always trace its learning process, it is, but we also don't fully understand the forces that render a "Scumak" sculpture into its shape and that's how we get to the question of how art made by AI can be compelling.

Let's assume for a moment that there is a work of art that we can accept as truly separate from human involvement.  There may be a lot to appreciate about it in terms of its formal qualities, optical effects, or synthesis of vast amounts of information, but when I look at a deep dream generated image, for example, the thing that compels me most is the algorithm that underlies it.  I linger over these images not to ponder how my mouth can become several dog eyes, but because I want to figure out which data sets it pulls from and who came up with the generator's optional layers and what assumptions about art were lodged in the biological brains of the human team behind it.

 (06:00) to (08:00)

We're led back to the age-old question of artist's intent.  Does it matter whether the artist is sentient or not, and is it even possible to judge a work without a consideration of its author, context, and origin?  If you're someone who likes to see evidence of a lot of hard work when you look at art, does it make a difference to know that a robot made it?  Would you find enjoyment in marveling over the kind of sustained, precise work an automated device can accomplish that a human probably couldn't, or would you rather marvel over artwork made by the humans who figured out the optics of some of our best loved works of Neo-Impressionism or Australian aboriginal dot paintings or even this painting, executed primarily by the assistance of Damien Hirst, who nevertheless has insisted, "Every single spot painting contains my eye, my hand, and my heart."

One of the reasons you might appreciate art, sentimental though it may be, is because it's a connection between you and another human being.  You may be separated by centuries or continents, but you're spending time with a piece of paper that someone else spent time with.  You're contemplating an arrangement of forms or an idea that once resided in the mind or presence of someone other than you. 

When we look at objects from times and places and cultures other than our own, there's always a lot we don't know.  Archaeologists and art historians seek to discover the motivations behind these objects, to enhance our understanding of what it is, why it is, and how we might find meaning it.  Part of our pleasure in these objects might be that mystery, exoticism, you might call it.  We may not know why it was made, but we can still find things to like, seeing it through the limited filter of our experience, and enjoy guessing at its origins.

Let's look at this Nkisi N'Kondi power figure, made in the 19th century in the area that is now the Democratic Republic of Congo and if you haven't studied it, I dare you not to think about who made it, how it was used, where it was kept, what's inside of it.  If you fall into the camp of those who think art should speak for itself, then what do we do with an object such as this?  

 (08:00) to (10:00)

Dismiss it because everything might not be evident at first glance?  Whether you're standing in front of a grand history painting whose story you don't remember well or a Hudson River School painting that seems to have no secrets to hide, there is always contextual information you don't have.  Even considering this 2002 video work by Cory Arcangel.  If you didn't grow up playing Super Mario Brothers, it just doesn't really make sense.

There is a backstory to everything.  Nothing is independent and everything is interrelated.  The result of a complex web of associations and actions and effects.  Whether or not you see it or want to see it is up to you.  One reason I like art is because I want to find out what motivated the person or people to do the thing, and when I wonder about the why of an art generated by AI, I'm either led back to a human person or to nothing.

We will see more and more art generated by AI, likely of increasing sophistication, and I mean, if machines and algorithms begin to have personalities and quirks, I might be interested in experiencing connection with them through the art they make, but in looking at AI generated work through the lens of art history, I see it as part of a long story of collaboration between humans and technology.  While the line between the two has become increasingly indistinct, I believe it's that murky territory that we find compelling experiences and ideas.  

What do you think about when you look at AI generated work? Does it matter to you if a machine or a person made it?  Let's talk about it in the comments.

The Art Assignment is funded in part by viewers like you through, a subscription-based platform that allows you to support creators you like in the form of a monthly donation.  Special thanks to our grand master of the arts, Indianapolis Homes Realty.  If you'd like to support the show, check out our page at


 (10:00) to (10:13)