Machines and People

Art is the medium that allows an artist to communicate something profound and meaningful to his or her audience in a way that does not merely inform but truly inspires...and which also allows the artist to transcend the brevity of human life to speak not solely to contemporaries, but to countless future generations as well.  When put that way, the underlying concept sounds fairly abstruse. But when considered in the context of real life, it feels almost natural: when we sit in the audience and watch King Lear talking to his daughters on stage as the curtain goes up and the play begins, it's not at all difficult to understand that it's only him talking to them in a certain sense, but—and far more profoundly—it’s really the playwright talking to us. Indeed, the difference between a great artist and a hack lies precisely in his or her ability to communicate deeply and movingly with an audience in a way that merely telling them that same information would not even slightly accomplish: what we learn in a few minutes of King Lear about parent-child relations and the degree to which greed can poison even the most natural kind of love couldn’t possibly ever be conveyed as deeply or as effectively by even the most talented university lecturer giving a public talk about the ins and outs of childrearing. Or about the nature of love. Or about greed.

That all being the case, art requires three things (or feels as though it must): an artist, an audience, and an artistic medium of some sort. The first and the second absent the third is just two (or more) people standing in a room. The first and third absent the second is the artistic version of a tree falling in a forest with no ear drum present to vibrate sympathetically when the tree hits the earth. The second and third is, at best, unrealized potential, a batter at the plate and a ball resting on the pitcher’s mound…but no pitcher in sight actually to throw the ball and, as such, no game to watch and either to enjoy or not to enjoy. And, of course, also no winner or loser.

So that’s two living, breathing people and one artistic medium that feel requisite. But now that we live in a new world in which machines can think—if not quite in the way human begins do, then at least to an extent that even a quarter century ago would have been unimaginable—the time may have come to revisit that those requirements.

Take, for example, these eyes:

They are expressive, thoughtful, fully human. It is a man or a woman? Is that the hint of a moustache under his nose or just a shadow? These eyes suggest a certain sadness to me, a certain world-weariness born of insight into the way that people are so often their own worst enemies. Without being able to see the rest of the face, this person seems to exist outside of time. If the rest of the picture depicted him or her dressed like an Italian aristocrat of the sixteenth century, I could believe it. But if the rest of the picture portrayed him as a cowboy or her as an astronaut, I could believe that too.

Here’s the rest:

So, not a cowboy or a doge, but a Dutchman. And this, I can hear you thinking, must surely be a work of Rembrandt, the greatest of all portrait painters and (of course) a Dutchman himself. But this painting is neither a Rembrandt nor a work by any of his contemporaries or students. It was created by a 3-D Printer that was programed over the course of an eighteen-month experiment by a team of art historians, computer scientists, and engineers brought together by Microsoft, the Delft University of Technology in Holland, and two Dutch art museums, the Mauritshuis in The Hague and the Rembrandt House Museum in Amsterdam. Bringing together digital data culled from 346 of Rembrandt’s real paintings created between 1632 and 1642, the idea was to create a portrait of a man not only dressed in the style of the time and with facial features similar to the men in Rembrandt’s real paintings, but to use the finest gradations of shading, texture, perspective, brush usage, pigmentation, and lighting to create a new portrait, one of no one at all but that surely feels as though it could be of someone whom Rembrandt could easily have known.

Is that art? It’s hard to say. The work has an audience and it exists…but does it have an artist? Clearly, a 3-D printer is not an artist, just a machine that does its programmers’ bidding. But are its programmers then the artists? I want to say no, that this project was just some digital silliness dreamt up by people because they had the technical skill to pull it off. But then I look again at the man’s eyes…and I feel a certain sense of kinship with this non-man who never existed. Does that make me a crazy person? Or does that make this a work of art?

Christie’s is about to auction off a portrait called “Edmond de Belamy, from La Famille de Belamy,” a work created by an algorithm (whatever that means exactly) and thus a product solely of its machine-creator’s artificial intelligence. The bidding is going to start at $10,000. The creators, if that’s the right word (since they specifically did not create the painting), are a trio of French businessmen with degrees in business and computer technology who call themselves Obvious. No artistic implement was used to create the picture—no pencils, no paints, and no drawing tools of any sort. Nor was human creativity involved other than tangentially: what the members of Obvious did, almost simply, was to feed thousands of portraits from the 14^th to the 20^th centuries into a computer that had been programmed to analyze the images in a dozen different ways and then attempt to mimic them as best it could. And here is, so to speak, Edmond de Bellamy himself:

Is this art? Most of me still wants to say no. But I find myself unexpectedly unsure as I look carefully at the painting and allow it to speak to me in precisely the way great works of art communicate outside of language and without being themselves animate.

I saw Her, Spike Jonze’s 2013 movie, and came away unconvinced that a man could truly love a machine, even one possessed of as intelligent and enticing an operating system as the one whose voice in the movie is Scarlett Johansson’s. Machines are not people. They cannot love. They cannot reproduce. But can they create? That is the question the portraits pasted in above awakens in me.

These questions lead to others. Can machines make music? Can they write books? Can they make scientific discoveries other than by processing huge amounts of data that their human masters have programmed into them? All these views have their proponents. Listen, for example, to Drew Silverstein, the CEO and co-founder of Amper, a company eponymously named after its sole product, an artificial-intelligence music composer.  Touted as the ultimate in artificial creativity, the program, so claims its founder, can create “unique, professional music tailored to any context in seconds” once you’ve provided it with the style of music you wish it to create, the mood you’d like to convey, and the length of the piece of music you wish to end up with. It’s beyond impressive. (To hear the whole spiel, click here.) And the product is certainly something like music. Maybe even it is music…at least in the sense that what they market as “cheese food” is some version of cheese. But what it lacks is the inner quality that, at least for me, defines what music—and what art itself—is: the ability to transcend the temporal and physical boundaries of the universe to communicate deeply moving ideas and emotions through the medium of human creativity. And that is what is lacking in all of the above. If there is no human artist, then there simply is no one for me to commune with through the medium of his or her art, no one to speak to me either deeply or superficially. Or at all. And without that psychic bridge between one human heart and another, all that’s left is technique and content.

Coming closer to my own turf, I find myself wondering if machines can write books. You may recall reading in George Orwell’s 1984 about a world in which the “proles” of a dystopian future solely read books written by machines. You may also be aware that amazon.com features over 10,000 books by one Phillip Parker, each of which is computer-generated and so, at least in some sense, “written” by a machine—but those books are merely compendia of facts and data, so hardly literary works other than in the sense that tax returns are or that telephone books would be if there still was any such thing. But other efforts are more intriguing. A Russian computer scientist, Alexander Prokopovitch, programmed a computer to produce his (or do I mean, its) 2008 novel, TrueLove, an attempt to tell the story of Tolstoy’s Anna Karenina in the style of one of my own favorite authors, Haruki Murakami. It was, however, not deemed a particularly successful undertaking and is no longer in print. (For a fairly dismal appraisal of Prokopovich’s efforts, click here.) Others will do better, I’m sure: to teach a computer to produce a text that retells a story that it has been programmed to regurgitate on command using a specific set of literary quirks and tendencies it has also been programmed to bring to bear in its effort to recast the story in different words doesn’t sound anywhere near impossible. But we’re back to the tree in the forest: if there is no beating heart inside an actual human breast with which I am being invited personally to commune through the medium of that person’s art, then there is—at best—a document, a story, or a book…but not literature. An image but not a painting. Sound, but not music. 

The bottom line, at least for me, is that art should be defined first and foremost as a mode of communication, as a way for two souls to meet even if their possessors never will or even could. If there is no other person involved, then even the most sophisticated effort to mimic art is just so much unrealized potential. Art, like love, requires two.