I was watching a Sunday morning news program recently when one particular story caught my attention. It was about using artificial intelligence to create artworks.
CBS correspondent David Pogue spoke to a software designer named Aditya Ramesh, who works for a company called OpenAI (a 2015 start-up of Elon Musk and others). Ramesh created an artificial intelligence program that creates original artwork in whatever style you like.
The software has been fed something like 600 million paintings, including everything from the Renaissance masters to unknown contemporary strivers, and it can draw from that vast pool to mimic any technique you desire. Just give it a subject and a style – say a girl on the beach in the style of Renoir – and it will paint you an “original” picture.
The program is called DALL-E, which is a play on the name of Salvador Dali and the Pixar robot WALL-E. Other AI programs performing a similar function include Midjourney and Stable Diffusion, and more are surely on the way. Nor is artificial intelligence confining itself to the visual arts. There are programs that compose music, produce poetry and write fiction.
These days, it seems like you can’t turn on a radio, TV or other medium without hearing something about artificial intelligence, but do you really understand what the term means? If not, you’re in good company, because I can’t find a clear and specific definition. Here’s what Oxford had to say about it: “The theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making and translation between languages.”
As you can see, there are holes in that definition. What, for instance, does “normally” mean in this context? I have found various other definitions, but all of them seem vague, subjective or overly broad. Basically, if a computer is doing something that strikes you as human, you can call it artificial intelligence.
Obviously, the whole field raises a raft of questions when it comes to the creation of artworks. The first one that comes to mind for many of us is what effect the programs might have on the employment opportunities of working and future professional artists. Will a lot of that work get sucked up by machines?
Those of us who care about the art materials market would certainly want to ask how these “paintings” would be rendered. Are they digital creations that only exist on electronic screens, and if so, are people expected to scroll through their art collections the way they do their Facebook messages? Are they going to hang flat-screens all over the house?
I also think that paintings would lose something without paint, and I’m not sure how AI would address that issue. Robot painters, perhaps. I’m picturing R2-D2 in a beret.
Thus far, professional artists themselves haven’t been too concerned about losing work, confident in their ability to produce something that cannot be replaced by virtual reality. What they are more concerned about is the likelihood that they could be robbed of their intellectual property rights.
Artificial intelligence applications require an enormous amount of data on an ongoing basis, and much of that data is coming from open source content on the Internet. “Open source,” however, does not mean the same thing as “free.” Users must sign an agreement which may prohibit the resale of code that is copied from the open site. In spite of that, artists are afraid that AI companies will ultimately make huge profits on software that is based on artwork of people who were not paid anything.
The same basic situation applies to AI writing applications that create articles, essays or whatever, written by computers in the style of whomever you want. The best-known player in that space is an app called ChatGPT, which also happens to be owned by OpenAI.
That company, along with Microsoft, GitHub, Midjourney and Stability AI, is currently the subject of class-action lawsuits that allege unauthorized use of proprietary materials. I won’t comment on those proceedings, partly because the technology is way over my head, and partly because I’d prefer not to become part of a lawsuit myself.
I am intrigued, nonetheless, by the basic question that underlies much of the debate about artificial intelligence: To what extent is it okay for one artist to copy the work of another?
You can consult a lawyer, but my understanding is that it is permissible to make a reproduction of a painting, so long as you don’t try to sell it as an original. That is called art forgery and is illegal. In terms of copying the style of another painter, there is nothing to stop you. You can paint an original landscape in the style of van Gogh and sell it, so long as you don’t sign his name to it or otherwise try to convince the buyer that he painted it.
I’m not a painter, but I am a writer, of sorts, and the same basic principle applies. In most cases plagiarism is not technically illegal but it can end a career or become part of a lawsuit. Writing in the style of someone else, on the other hand, is usually acceptable and sometimes even considered laudatory or amusing.
When I’m reading for pleasure I tend to gravitate toward detective fiction, where copying styles, along with specific characters, has become a genre unto itself. I recently read a new Sherlock Holmes novel, despite the fact that Holmes’ creator, Sir Arthur Conan Doyle, died nearly a century ago. The author of the new book has written 23 Sherlock Holmes novels – way more than Doyle himself – and dozens of other writers have written hundreds more.
In fact, Sherlock Holmes is the most portrayed fictional character on stage and screen, and most of those productions have little to do with Doyle’s work. As far as style goes, however, there is one writer who is copied more than Doyle or anyone else.
Raymond Chandler virtually invented the sort of noir, hard-bitten private detective that became such a cliché in post-war American fiction. Chandler’s hero, Philip Marlowe, has served as the model for countless imitations over the years, and has been portrayed in films by such distinguished tough-guys as Humphrey Bogart and Robert Mitchum.
Every year, there are contests to see who can write the best story in the style of Raymond Chandler, but of course most of the contestants exaggerate to the point of caricature. The novelist Robert B. Parker, a very successful detective writer in his own right, went so far as to finish a book that Chandler had started 40 years earlier.
I read it, and I have read the work of many popular authors who were heavily influenced by Chandler. Some of them are very good and worth reading on their own merit, but in my opinion, nobody has ever quite captured the essence that made the original so compelling.
Technology is constantly and ever more rapidly improving, and I have no doubt that artificial intelligence will gradually take over some of the more mundane facets of art creation. In other words, they will become the tools used by the artists of tomorrow.
The soul of artwork, I believe, will remain a very human function.
You can e-mail Kevin at email@example.com