Uncanny Valley: Being Human in the Age of AI

Today's post is somewhat off-topic, despite having AI in the title. Uncanny Valley: Being Human in the Age of AI is the name of an exhibition at the de Young Museum in Golden Gate Park in San Francisco. I went to see it before we got locked down. Of course, the de Young Museum is closed to visitors at present. But this exhibition is scheduled to run until October 25, so I assume there is a chance that you will be able to see it if you want.

In case you don't know the phrase "uncanny valley", it has to do with computer graphics. People are quite happy to look at obviously artificial characters such as Disney or Pixar animated movies. They are also happy watching anything where the graphical rendering is so good that you can't really tell it is not real. But in the middle is the "uncanny valley" where the modeling and graphics is almost good enough but not quite. This makes people feel uncomfortable. The most well-known movie that fell into the uncanny valley was Polar Express—if you've seen it you probably found it eerie to watch.

I first heard about the exhibition since I ran into a story about how one of the artists from overseas had been "denied a visa by an AI algorithm". However, I'm a big believer in Hanlon's Razor:

Never attribute to malice that which is adequately explained by incompetence.

Over the last few years, I've seen several people complain about having visas refused. But in all cases, what happened was that they applied, and then never heard back from what I still think of as the INS but these days is, in fact, the USCIS. They never get refused, and the visa probably got approved (too late to be useful). I've had that experience myself.

Anyway, this is a post about an exhibition, not immigration, but the description of someone not getting their visa approved or denied as "due to an AI algorithm" seemed to be very typical of the understanding of AI: a mixture of technical confusion along with positioning everything in the most anti-establishment way possible.

Simon Denny

I am also not sure how much the cards on the wall that accompany each piece really reflect the artist's views, as opposed to the curators of the exhibition. I've seen complaints before by artists that the explanation of what the art is "about" is simply wrong, and just reflects the curator's prejudices. The funniest example of something like this, although I can't find it right this minute, was an author on Wikipedia who corrected some misunderstanding about the plot of one of his books, only to have the corrections first marked as ^{citation needed} and then they were reversed out. The fact that he was the author didn't count for anything against citable papers from "experts".

One good example of a card like that is Simon Denny. Here's the opening of the card describing his exhibit:

Simon Denny conceived the sculptures and collages here partly in response to a 2018 essay on the Echo, Amazon's home surveillance artificial intelligence system, cleverly marketed as the personal assistant Alexa. In their essay, "Anatomy of an AI System: The Amazon Echo as an Anatomical Map of Human Labor, Data, and Planetary Resources", Kate Crawford and Vladen Joler trace the extractive processes involved in the Echo's production. They expose how its sleek design belies a trail of environmental disruption in its wake. Connecting physical mining to data mining, it suggests that our trade in data is just as detrimental to our planetary health.

In the grand scheme of things, there are just not that many Echo devices, maybe 100M, and their environmental impact has to be tiny compared to power generation, automobiles, or even smartphones. And lots of us seem to be happy to have a "surveillance system" in our homes.

Agnieska Kurant

Interesting to see were Agnieszka Kurant's termite mounds (and by the way, her name is spelled differently on the card on the wall and on the museum website). Since the theme of the exhibition was humanity and AI, it was a bit of a stretch to say that:

By repurposing the colony's collective intelligence as art production, the artist comments on the erosion of singular authorship.
...
Platforms like Amazon's Mechanical Turk allow individuals and corporations to post computer-based tasks that the workforce carries out. The worker's invisibility creates the illusion that the tasks are being created by machine, leading some to describe their labor as "artificial artificial intelligence".

Zach Blas

This piece is The Doors, a "six-channel video projected onto glass panels set in an artificial garden populated with fake plants...opening a portal into tech culture, it highlights the industry's conflation of 1960s countercultural attitudes with neoliberal ideals".

I've worked in tech all my life, including in several startups, but I must go to all the wrong parties:

Entrepreneurial startups, once thought to be grounded in countercultural and libertarian ideas, have been transformed into capitalistic megacorporations that use the psychedelic experience to optimize brains and bodies for labor.

Forensic Architecture

The most dramatic exhibit was the huge wall of faces down the entire side of a gallery. The exhibition website describes this as:

"They Took the Faces from the Accused and the Dead" consists of a large gridded installation of more than three thousand mugshots from the archives of the American National Standards Institute. The institute’s collections of such images were used to train early facial-recognition technologies—without the consent of those pictured.

Of course, my immediate thought was to wonder about the philosophical difference between neural network training or creating a wall in a major exhibition.

Also by Forensic Architecture was my favorite exhibit, a large number of painted and photographed shells (the military kind, not the seaside kind). Actually, they are 3D printed with "with Polyjet technology in full color".

More About the Exhibition

There was a lot more than just the few pieces I have managed to highlight here. It is an interesting exhibition. If you are reading this blog post, I doubt that you will learn anything about AI from a visit. But it is interesting to see how people from outside technology view it looking in.

Martine Paris of Forbes magazine said, about this exhibition:

Uncanny Valley will leave you questioning just how far we want to go down the rabbit hole.

Well, if society opens up again, you can go and see how deep you want to to. As I said earlier, the exhibition is until October 25. It has its own page on the de Young Museum's website. Or, if you decide you want to go really deep, you can purchase the 224-page exhibition catalog for $45.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.