Painting By Numbers


Robots and artificial intelligence (AI) are often thought of as basically the same thing, probably because science fiction movies and TV shows have portrayed them as such for the past half-century or so. If I say, “robot,” you probably visualize a very smart machine that can both walk and talk. It may not look like us, but it likely has two legs, two arms and a head.

Such things do exist, at least in a primitive fashion, but technically robotics and AI are two different fields. Robots are machines that mimic a physical human function, like an arm that applies welds on an auto assembly line. Artificial intelligence is the use of computing power to mimic human cognitive ability, like the IBM machine that plays chess.

Lately there has been a great deal of speculation about the extent to which AI can go. Is there a point where computers will reach something akin to consciousness, and will start to feel genuine emotions? If so it seems logical that they might develop agendas of their own.

That may sound silly to you, but it’s a serious matter to futurists such as Elon Musk and scientists such as the late Stephen Hawking. Both have said that AI is the greatest threat to the long-term survival of our species, and that we might want to rethink the whole idea of combining intelligence with mobility.

It may be a while yet before they come for us, but robots are already having a major effect on our economy. We spend a lot of time and energy worrying about American jobs that have migrated to other countries, but economists estimate that 80 to 90 percent of the jobs that have been lost in this country over the past several decades actually went to robots and other advances in technology.

You may not think of the retail business suffering job losses from anything other than store closings, at least in terms of sales associates. It’s not like Americans are shopping that much in foreign stores, and not many retailers have fleets of robots patrolling the sales floor. (There have been some pilot programs.)

Retail, however, includes such mass-market outlets as fast-food restaurants, gas stations and convenience stores. Those places have been quietly moving towards automation for a long time, and that process will probably accelerate now that a lot of states are pushing to increase the minimum wage.

Sometimes the changes are not readily apparent to customers. When a voice at the drive-up window asks if you want fries with that, do you really know whether or not it is connected to a human?

The changes are more obvious in many supermarkets and big-box department stores, which have combined self-service with technology to speed checkout, while at the same time reducing staff. At the Walmart in my town, there are very few employee-manned cash registers open these days.

I have always thought that one of Walmart’s weaknesses was that you couldn’t find any sales help when you need it, and it would seem as though the self-checkout would only make that problem worse. I half expect the automated cashier to ask me if I found everything I was looking for.

It’s hard to imagine this level of automation in an art supply store, let alone the idea of robot sales associates. The image of knowledgeable, human, customer assistance is so deeply entwined with my concept of the art materials retailer that I can’t see the industry without it.

Then again, I couldn’t believe that self-driving cars was really going to be a thing, but it is. There are thousands of them on the roads already, and that is going to ramp up into the millions in just the next few years. The automakers have poured billions into development costs, so there will be no turning back. You can add cab drivers to the list of job casualties.

Whether your employees go robot or not, there is a larger group in play that is even more critical to your business – your customers. There is a very real debate going on right now over the issue of artificial intelligence being used to create art.

Last February, David Pogue wrote an article for Scientific American magazine entitled, “The Robotic Artist Problem.” In it he argued that the common notion that art would remain the “last bastion of human exclusivity” was not necessarily correct. In music and in painting, highly sophisticated programs have been developed and implemented, allowing computers to design original works in seconds.

At Rutgers University, for example, the Art and Artificial Intelligence Laboratory, led by Ahmed Elgammal, has been generating paintings in large numbers. Team members select style, genre, colors, subjects, or whatever, click on “render,” and voilà. Pogue refers to the resulting canvases as “beautiful, polished, and appealing,” but is it art?

I had a college professor who taught us that art was whatever an artist said it was, and an artist was whoever claimed to be one. It was simply up to the rest of us to determine whether or not the art was any good.

Computers have not yet reached what most of us would call consciousness, so I’m not sure that they can declare themselves to be artists, but a lot of people don’t really care. There are businesses that buy artwork in bulk, and they are far more interested in cost than in provenance.

The definition of art aside, Pogue asks the question he seems to find more relevant, which is how do we put a valuation on art that is generated by artificial intelligence. His answer, at least in part, is that value is determined by scarcity. The reason that the Mona Lisa would be more valuable than 100 exact copies is that there is only one original, so theoretically AI artwork could also be valuable so long as each piece is unique.
There are a lot of holes in that argument. One of them is pointed out by an article in The New York Times from July 8th, “Do You Like ‘Dogs Playing Poker’? Science Would Like to Know Why.”

It reports on a study completed this summer at Boston College which uses quantifiable data and statistical analysis in an effort to solve that old question of why we like what we like. The study determined that one of the primary factors was people’s belief that the artist leaves some essence of him or herself behind. I don’t know what sort of essence is left on an AI canvas, perhaps some echo of the original programmer.

The researchers did discover one thing that I thought was pretty funny. Apparently people like an AI artwork better when they see a humanoid robot painting it.

One question was not addressed by the study, and I think we should look into it. “How do robots buy art supplies?”

Leave a Reply

Your email address will not be published. Required fields are marked *