picture of ro

ChatGPT and other “generative AI” chatbots are causing disruption in many areas. You’ve probably read a lot about student essays written by machines and then passed off as the student’s own work. You may not have read so much about journalists and bloggers using the same tools to generate news copy. Then there are other professionals who have to write things – business letters, reports, press releases, memos – using generative AI to create polished prose on demand. What you may not have heard about is a growing trend for actual writers – creative writers – to use the tools to help them write short stories, poetry, even novels.

This recently became a minor sensation among my fellow authors and editors when a sci-fi and fantasy magazine called Clarkesworld stopped accepting submissions because it had suddenly become swamped by AI-generated stories. The view of the editor-in-chief is that this is an illegitimate activity perpetrated by scammers. In the wider world of writers and editors, there is grave concern, not that the stories will ever be published – generative AI isn’t that good, yet – but that the ease and speed with which such stories can be produced, means that many or all literary magazines will soon be unable to function because the system will be clogged up with dross that will drown out the voices of real writers.

It’s a reasonable concern, I suppose, although a bit of computer-assisted pre-processing ought to be able to weed out the vast majority of AI-written submissions and restore the whole submissions review process to its former stately progress. But at the moment, it is disruptive.

Perhaps a more important concern is that some people believe it isn’t appropriate for machines even to attempt to produce art. Leaving aside the sad fact that most published writing never approaches the level of what might be called “art” and is best described as “entertainment” (indeed, I suspect that most writers, like me, don’t even aspire to write anything more profound that a few good ideas, an enjoyable story and some interesting characters), and that most readers are not in the least interested in reading “art”, the idea that “art” is and can only ever be a human activity is an interesting one.

“Machines can never be creative” people are saying. “Art is a conversation between a writer and their audience,” they add. It’s never clearly stated but I suppose they believe there is some kind of divine “spark” within the human soul from which “art” comes and that “art” can never come from any other place. To me, this is a strange and essentially religious position to take.

As someone who was involved in AI research in the eighties and nineties and has retained an interest ever since, I have read much by the people who have devoted their lives to the task of creating machine intelligence. They have grappled deeply with ideas like “creativity” for decades and have sought clarity and precise definitions that might one day become the basis for automation. They have also dug deeply into ideas like “understanding”, “experience”, “explanation”, “knowing” and so on. Many AI researchers are basically philosophers with an urge to reify these difficult, “open textured” concepts. It’s why I found the field so very interesting and have such high regard for the people working in it. It is also why I believe machines will one day be writing brilliant, moving, exciting and insightful novels – at least at the level of most published human authors. It’s why I think that “art” is not just a human preserve.

Of course, in the short term, people using chatbots to write or help them write a short story every ten minutes, or five blog posts a day, or to pack the tabloids and trade magazines with column-inches of bland, mediocre text, will be the ones pushing the technology out into the world and making editors miserable. However, in the longer term, it is the scientists and engineers working on semantic models and emotional intelligence, discourse structure and creative insight, who will move the technology so far past where it currently sits, that, if editors want to exclude machines from their periodicals and booklists, they will need to find some other criterion than literary quality.

2 thoughts on “Writers vs Generative Artificial Intelligence”

    1. Thanks, Rod. Loads of interesting material in there. Interestingly, it is all written from a very positive, even enthusiastic standpoint which, I must say, is where my own sympathies lie. Yes, there are IP issues, especially with artworks (and the discussion on “derivative works” in IP law was actually very helpful for me) but the immense usefulness of these tools will probably force us to confront and solve such issues before long.
      As you might have noticed, I used generative AI to create the artwork for my novel, “Metaman” (as described in this post: https://www.cantalibre.com/blog/2022/08/15/using-ai-to-create-cover-art/). I probably dudded some artists out of some royalties in doing so but I’m hoping that, in the future, using such tools as DALL-E might involve charges that are routed to the original artists (if significant contributors to the model can be identified and there are not hundreds or thousands of them).

Leave a Reply

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