We’re living through one of the greatest changes to civilization since the industrial revolution. Technological infiltration into almost every aspect of society is faster and more pervasive than many people realize. Computers will continue to make our lives simpler by relieving us of repetitive and mundane tasks while improving efficiency. However, a question that often arises is, how far will they go? Are we now, for example at the threshold of computers creating legitimate art? Some experts are suggesting it’s not only possible, but will happen sooner than you might think.
What follows are what I believe some pertinent discussion points as this topic relates to creative writing. To start, I think it is fruitless to argue that computers won’t be able to write something that resembles your favorite beach read … someday. They can create a simple text now and that capability will only increase as time goes by. However, just because they can doesn’t mean they will. They can’t get there on their own (at least, not yet!) A human must decide it would be a good idea for Robot Writers to replace the Grisham’s, Connelly’s and Patterson’s of the world.
“Computers can’t write! They only follow instructions.
There is writing, and there is writing. Did you know, for example, that the Washington Post’s computer reporter wrote 850 articles in 2017? I found an estimate that Google robots wrote ten percent of all Gmail’s in 2018. When you read the morning blurb about your favorite sports team, chances are a computer produced it. It would appear that the programmers’ imagination and code-writing skills are the only limits to which computers will take over writing tasks.
But creative writing is different! (Right?)
Yes and no. When I first started learning the trade from professionals close to fifteen years ago (I’m still a student and always will be,) I remember my surprise at the level of structure embedded in my lessons. My teachers at the University of Toronto insisted that every scene, indeed the entire story must contain seven attributes. These included an opening line hook, an inciting incident or surprise, an overarching dramatic question, escalating conflict, a reversal or revelation, a climax and a controlling idea. If you Google “The Hero’s Journey,” (another well-known writing road map) you’ll find a similar regiment. These guidelines intend to keep the manuscript both organized, entertaining and thus easier to read. We’ve all closed disorganized and hard-to-follow novels long before the final page. (The first rule is, “Don’t lose the reader!”) Good writing is thus fashioned with the reader in mind and contains a prescribed equation for success. The irony, however is that very structure is what lends itself to automation.
Computer software already assists in the novel creation process because writing is much more than recording a series of epiphanies. Editing takes up many more hours than completing the first draft. A quick internet search revealed no fewer that a dozen programs that will check your writing for things like clichés, poor grammar, overused words and clunky sentences. It will highlight what part of The Hero’s Journey you are missing or is out of place, will help you past the dreaded “writer’s block,” and assist you in building characters by offering hundreds of choices. (Think Dr. Frankenstein’s monster.)
Could a computer “learn” to write novels from scratch?
We need to consider machine learning, which is defined as, “The development of computers to do tasks, that if done by a person would require intelligence.” These include things such as reading, writing and learning from mistakes. I’m sure you’ve noticed when you compose a text or an email, the program tries to finish your sentences. When you don’t agree with the auto-suggestion, it takes note. Over time, the program’s accuracy will improve because it will learn from its “mistakes.” The more emails you write, the greater the correctness.
This is the essence of Artificial Intelligence (AI,) and if you find you’re hearing more about AI every day, there’s a reason. New Vantage Partners’ 2018 annual executive survey found 97% of executives representing almost sixty Fortune Financial companies were investing in AI. It is here to stay… and grow. Many people I suspect, including me, don’t understand all aspects of it. If you want to give your head a bit of a spin, look up “Neural Networks and AI” and have a read. (Hint, they are designing AI to transfer information in the same way as the human brain.)
The “learning” happens through data collection and analysis and that is key. I believe this might be the least understood part of today’s revolution. I’m sure you’ve heard the phrase, “It’s all about the data.” Those in the know describe the size of the world’s data in thousand’s of exabytes, a number beyond my comprehension. Many of us would say, “They’ll never be able to build a computer capable of handling that.” The fact is, in a cloud computing world where lots of processers work together, the capture and the manipulation of huge amounts of data (to the benefit of the operator) is straightforward. This means the machines learn quick. For example, the latest chess-playing computer program called AlphaZero taught itself the game from scratch in four hours and is now contending for the title of world champion chess program. It only started with the game rules and the aim not to lose. It developed by playing against itself and “learning” from its less than perfect move selections. I’m thinking multiple millions of moves, if not more.
Let’s say you wanted to “teach” a computer to write a fiction novel. First, you might create your database by downloading every word of every such genre book ever written. Depending on who you believe, the US Library of Congress holds between ten and twenty terabytes of some twenty million books, and I found more than one prognosticator that said it would take about fifty terabytes to store all fiction ever written. But the key is, computers can search through this data to find whatever you tell them to find. Remember, computers “create” by processing huge amounts of data, coming up with the “best” answer and then finding an even “better” one.
Is calculating the past (to a mind-boggling level) the same as ingenuity?
This is an important question. Now that we’ve downloaded every book ever written can the program generate new ideas, slants or approaches for telling a story? The contest is between the computer’s ability to analyze and learn from what already occurred vs. the human mind’s apparent unlimited capacity to create. Think about what it means for a novel to be “good.” All fiction readers know the feeling of getting “lost” in a story, losing track of time and perhaps even where they sit as they turn page after page. They want to dream with the author and that is every fiction writer’s ultimate aim. A great book captures a good part of the present. I received an email from Barbara Kyle, a celebrated author and teacher of mine. She reminded all on her list that “agents and publishers want… page-turners. A book that leaves the reader saying, ‘I couldn’t put it down.’” She emphasized character creation, story structure and style, but finished with the goal of creating a “meaningful emotional experience” for the reader. You can program a computer to enhance character, story and style, but it is this last point, to me at least, that separates man from a machine. Author’s represent the entire gambit of human emotion. The words we choose, whether we realize it reflect our entire histories. We each observe, absorb and reflect on life’s events in our own unique way. Taking this a step further, sometimes the words we write don’t mean only what they appear to. Countless authors imbed a double entendre in their work. Shakespeare was famous for it. In Robert Frost’s poem called Birches, a surface read depicts a fascination with young boys climbing up birch trees to swing them. Closer scrutiny, however reveals Frost’s fear that climbing trees meant you were getting closer to heaven. He was a very religious man. The boys must be careful to never get too close because that would mean they were in heaven and no longer on earth. Can a computer do this? We’ll have to see. AI developers will tell you that the machines are showing curiosity and creativity. I have to assume the coders can get their machines darn close to creativity. Never say never.
But, getting back to my main point, someone has to make it happen.
Why would anyone want this?
Put another way, who stands to gain? That’s important from a business perspective. Like anything else, computer development takes time and money. The three potential benefactors include the authors, the readers and the publishers (meaning all the people who profit by someone else’s creation of the fictional story, including a book, TV and movie/videos.)
I’ll start with the readers. There are huge numbers of books written every year that don’t get read. Introducing computer written novels would increase that oversupply overnight. If I remember my economics 101 course, increasing supply doesn’t on its own affect demand. What happens is the price goes down. Within reason, I believe the demand for books is more price inelastic. (They always seem on sale anyway.)
If it isn’t lower prices, what else will spur demand? After thinking it through, I don’t see readers preferring computer written novels. Much time would have to pass before consumers considered them a better product than what a person produces. There’d be some amount of resistance to the idea at all. How many readers wonder what an author was thinking as they built their story? There is a “relationship” of sorts between a human writer and reader. Could they get the same feeling from a book they knew a machine wrote?
How about the publishers? Is R2D2 likely to replace Grisham as a cost saving move? In very rough numbers, an author will receive from 5% to 7.5% of the book price (higher if hardback.) For example, on a book priced at $20, the author receives between $1.00 and $1.50. This seems small, but when sales volume hits the hundred-thousands or millions, the number is meaningful. A publisher should then want a computer to write their best-selling books, because that’s where they have to pay millions in royalties. The problem with that is, the author’s name, backed by something like “Oprah’s top pick” or “New York Times best seller” are key selling points. Admit it. We have at one time or another bought Grisham’s next book because it was Grisham’s next book. I don’t see the publishers’ motivation to bite the hand that feeds them. While this discussion is far more complex than I’ve described, I will assume that it is not the publishers that will push for computers writing their products.
That leaves the authors. First glance would conclude they are the last group wanting a new technology that would strip them of their jobs. But, there may be an Achilles heal. Going back to my chess analogy, I play against people from all over the world, hosted by a website with over thirty million members. Mostly, it is fair play, but you can today download an APP to your phone that will tell you your next best move. It’s better known as cheating and the site’s host’s boast about sophisticated algorithms they use to ferret out and expel the offenders just by studying their moves. There’s that dark side to human nature of some people wanting to win at any cost (including their honor.) Bring that scenario back to writing. While the exclusive group of world-famous writers won’t have their desktop transcribe their books, thank you very much, what about the ninety-plus percent of other writers? This is the market for the software programs, “better” versions likely to hit the market every year like Titleist golf balls. At what point does relying on software to write your novel become cheating? And, if a writer and computer can produce a page-turner, selling a million copies, to whom will it matter? I’m not arguing this will occur, but it is an open door.
It’s anyone’s guess how far and fast technology will advance. I know scientists are studying how the human brain generates creative thoughts. That sounded far-fetched to me until I read the other day that they have identified that part of the brain that creates speech. By hooking up wires to your head, someone who lost their ability to talk (e.g. through a stroke or some other malady) could think a message that a recipient could read, understand and then act on. How about, “Siri, turn the TV on to channel 13” when not a sound was heard? This isn’t science fiction. It’s a wonderful, but also scary hint of our future.
“Toto, I’ve a feeling we’re not in Kansas anymore.”
So, the answer to my opening question is “perhaps.” Maybe this is all a generational thing, each older one resisting change they no more want than understand. In my heart I believe, regardless of the medium, stories will still need creation by someone who spends an inordinate amount of time writing draft after draft, agonizing over each word they produced, laying awake at night wondering how their protagonist can ever prevail. Computers will help polish them, but the authors will still chip away at the slab of marble until the statue is complete.
That is at least until the human mind directly links to computers. That would be a whole new kettle of fish. (Oops, my editing software just warned me that’s an old cliché.)