One of the big topics in the writing world lately has been AI. There’s an AI program that supposedly can write as well as a human, given a prompt. And it’s already become a huge problem. Some science fiction magazines have had to close submissions because they were getting spammed with AI-written stories and couldn’t weed through all the submissions. Apparently, there are people who seem to think that selling short stories they don’t have to write is a get-rich-quick scheme. Give the AI some prompts, churn out a story in seconds, and then flood the publications with submissions, and something is sure to sell. It’s a quick and easy few hundred bucks made in minutes of work. Of course, that assumes that any of these stories sell, which they probably won’t, but there’s no loss to the submitter if they don’t, since they put no effort in. It’s the same principle as with spam e-mail. Might as well flood the world with it because it doesn’t cost anything more.

Except these AI submissions are pretty obvious and not that great, but the editors still have to go through all the submissions to find stories they might like. It’s hard to come up with a filter to screen out the spam that doesn’t discriminate against possible real authors who are just new and learning and don’t deserve a lifetime ban on submissions the way the people gaming the system do.

And, of course, there are already people “writing” novels with AI and publishing them on Amazon. A few minutes of work, then you’ve got a novel, and if it sells more than a few copies, you’ve still made a decent hourly wage. Even if most readers figure out it isn’t very good, it still increases the flood of books out there, making it harder to dig in and find the good ones.

I suspect that no matter how good AI gets at writing, AI-written fiction will never be that good because it comes down to the ideas. Just about any writer has had the experience of meeting someone who says they want to be a writer, but they don’t really want to write. What they have is a brilliant idea they want to share with a writer, have the writer write it, and they’ll split the profits. And I don’t know that I’ve ever heard a decent idea from these people. About 95 percent of them don’t actually have a story idea. They just think that their own lives are interesting enough to be a novel, so what they want is to write about themselves. I’m picturing all these AI-written books about boring men who’ve overcome some minor adversity.

What real writers know is that the idea is the easy part. Most of us come up with new ones every day. The trick is to figure out which of those ideas has what it takes to develop into a story and then to develop it so that it has enough substance to build a story around, and to do that, you have to write. Every time I’ve had an idea that feels totally complete, like I could just sit down and write the book, once I do start writing down everything I know about the story, I’m lucky if I get two pages. The idea is nothing without a lot of work. Feed that initial idea into a computer, and you’re not going to get any kind of decent result, especially since this is just a sophisticated auto-complete based on other people’s work. It merely guesses what word should come next based on all the input it’s received.

I think what a lot of the people using this tool want is to be published. They don’t really want to write. They just want a shortcut to the result.

Supposedly this tool does write good book descriptions, like what would go on the back of a paperback, but it can also be badly, hilariously wrong. I’ve seen one person describe it as more like having an enthusiastic intern who needs a lot of direction and supervision than like hiring a professional writer.