Monday, September 12, 2022

DO AI ARTISTS DREAM OF ELECTRIC SHEEP?

 Sigmund Freud claimed there have been three great shocks to the human ego:

  1. The discovery that the earth is not the center of the universe but is instead a tiny, vulnerable planet adrift in a cold and inhospitable universe.
  2. The discovery that humans descended from "a hairy, tailed quadruped" rather than being the divinely appointed progeny of Adam and Eve.   
  3. The discovery that humans are not sublimely rational creatures, but are instead controlled to a disturbing degree by our unconscious and the spasms of our residual lizard brains
Each of these three great shocks diminished our concept of humanity.  They also challenged us to re-define our species in a manner consistent with the new reality (reality defined as "that which, when you don't believe in it, doesn't go away.") 

To put it mildly, humanity has not responded well to these challenges.  Angry disbelievers wage violent rearguard wars.  In 1925 the famous Scopes monkey trial in Tennessee convicted a teacher for the crime of teaching evolution.  The judge refused to permit scientists to testify at the trial, ruling that scientific evidence would "shed no light."  It took more than 40 years before the US Supreme Court overturned that prohibition.  Nearly 100 years later, law and science remain no match for the resilience of disbelief:

Tennessee farmer O.W. Wooden: "Trying to tell you people come from monkeys
and all that stuff.  Couldn't be right!  Monkeys to me, like a chicken, you know?"

Today's question is whether it is time to add a fourth great shock to Freud's list: has artificial intelligence advanced to the stage where it is time to rethink our unique human ability to create art? 

Machine art has been rumbling like distant thunder for a long time.  Photography, moving pictures, reproduction technology and Photoshop have all transformed our historical concept of art.  They've reduced the difficulty (and therefore the mystery and value) of human skill.  They've significantly altered some of the fundamental goals of art, putting an emphasis on concept rather than execution, substituting curation of images for creation of images (using a whole new vocabulary such as "appropriation art," "re-contextualization," "sampling" and "augmentation.").  "Photo-illustration" has swallowed up whole categories of work that was once profitable for artists.  Digital art, which can be infinitely reproduced in perfect copies, has attempted to restore uniqueness and authenticity with the artifice of NFTs.  

None of these changes to traditional art would have occurred if traditional art wasn't forced to redefine itself in reaction to the boarding house reach of machine art.

Now artificial intelligence is rattling our door. 

In the past year, affordable off-the-shelf AI software (such as MidJourney and DALL-E2 ) has empowered adolescents with no discernible talent to create images using words alone.  If you know how to spell your illustration assignment, you can receive an offering of customized solutions within seconds.  As one self-styled "AI artist" exulted: "I felt so liberated bc drawing is the one creative thing I can't do AT ALL but I have a lot of hyper specific art ideas...." 


You can hear the "Yipeee!" echoing all across the internet. 

Discussions about the significance of AI art now abound, including in the comment section of my last blog post.  Predictably, a lot of the traffic on social media is precipitous as well as factually, legally and economically misinformed.  Still it's not too early to grapple with the question: have we finally arrived at Freud's 4th great shock?  If we attempt to deny it, are we any different from O.W. Wooden?  And perhaps most importantly, if the 4th shock is indeed here, how can we define our creative species in a manner consistent with the new reality?

There is no single answer because art is not a single discipline.  The shock waves from AI art will affect multiple artistic categories (economic, legal, educational, professional, and yes, creative) differently.

I think it's a mistake to take too much comfort from the fact that people with taste are still able to distinguish AI art from high quality human art.  Any serious ontological analysis must take into consideration the growth trajectory of AI, and search for qualitative barriers that AI won't be able to cross in the future.  

Pixar's ungainly CGI experiment, Tin Toy , came out in 1988. A mere seven years later Toy Story was released as a persuasive, full length CGI feature.