The Bot Did It: Should We Fear the AI Revolution? (by Pat Black)

Glasgow, Scotland native Pat Black returns to EQMM in our current issue, July/August 2023, with “Twos on That,” a thrilling entry in his Lomond and Slater series. A journalist currently living in West Yorkshire, Pat is also the author of six novels (as P.R. Black). In this post he takes up a topic that is on the minds of most authors and publishers these days. In addition to giving us some food for thought with this post, he has recorded “Twos on That” for our July first podcast. Don’t miss it!  —Janet Hutchings

Well . . . “yes,” is the answer to that one.

By now we’ve all had a chance to think about AI and its effect on writing fiction. ChatGPT and other AI-driven software has opened new frontiers in creativity—none of which augur well for the creative individual. Especially if that creative individual wants to . . . (hits CTRL plus I) get paid for it.

We’ve heard the counter-arguments already. That AI isn’t sophisticated enough. That it’s a jumped-up version of auto-correct, or the little paperclip person that helped you out on primeval versions of MS Office. I get that—but you’d be naïve to think that the technology will stay that primitive, that it won’t become more intelligent, more intuitive, even sophisticated.

I once mentioned in conversation that AI-driven technology might be a godsend when it comes to human medicine and surgery; where humans can err, and sometimes catastrophically, a machine may be much less likely to do so. I was mocked for this. And I concede: No, the technology isn’t there yet. But it will be one day.

I’ve no wish to discuss general doomsday scenarios as a result of AI’s interference here: There’s a lot of that speculation around at the moment, by people much better informed on the subject than me. But the technology and its possible use does provide a problem for the arts.

If we treat any story in any medium as a formula, whether in derivation from a strict algorithmic sequence or as a mutation of it, then it stands to reason that AI will replicate that formula and produce art. It won’t be good art, to start with, but like all computer technology, it’ll evolve, and improve.

Crime novels are perhaps the most formulaic of all, and therefore one of the easiest to replicate in this near-future of AI-driven authorship.

Take the whodunnit. You have the crime, usually murder. You have a victim. You have suspects. You have motives. You have clues, red herrings, plot twists. You have a resolution—on the author’s part, hopefully it’s one that readers won’t guess easily. You have a protagonist, too—in most cases, a detective. Add in a setting, both physical, geographical and temporal, and a method of murder, and, yep . . . .you’ve got it. It can be argued that all you need to do is fill in some blanks.

The key test will be: Can machines takes this formula and its possible permutations and make better stories than humans?

Part of the AI revolution is already taking place. At time of writing, the “Abba Voyage” is still ongoing in London, still packing in the crowds and becoming a tourist destination as much as the London Eye or the Houses of Parliament or Madame Tussauds. Although members of Abba did pop up at the venue for a record-setting performance the other day, there has been surprisingly little commentary over the fact that anyone going there is not actually watching Abba.

For all the chat about motion capture and 3D holographic smoke and mirrors effects and faithfully reproduced live vocal performances, anyone going to Abba Voyager is effectively watching a video. The “experience” part of it probably depends on your own personal Abba threshold.

My Abba tolerance has risen in recent years—everybody ends up liking them eventually—but I can’t help but feel that the impressive 3D “Abbatars” have an “Emperor’s new clothes” element to them. Some people have pointed out his bare bum, but not many. It’s a simulation. No matter how you dress it up, you’re watching a recording. It’s not a live event. The art is driven by machines.

I can foresee a future—not too distant, either—when AI linked to this no doubt closely-watched entertainment phenomenon brings back musical acts which are becoming rarer as their natural record reaches the end of the groove. Many outrages and blasphemies are possible, here.

Imagine The Beatles, reformed, down to the flecks in the black and white footage, playing brand new songs modeled by AI that sound as good, if not better than the classic Lennon/McCartney numbers. Any number of resurrections might take place—take your pick, really—Johann Sebastian Bach, Wagner, Elvis, John Lennon, Frank Zappa, you name it. All fully customisable. Let’s have Jimi Hendrix having a duel with Paganini, why not?

At the touch of a button, you might even generate collaborations, no matter how outre. Iron Maiden and Nana Mouskouri. Napalm Death and Conway Twitty. Lou Reed and Metallica . . . wait, that one’s been done in real life.

There are only so many musical notes, discernible to the human ear. There are only so many ways they can be combined, only so many chord progressions, only so many melodies, only so many beats. To a computer powerful enough, replicating these mathematical patterns might seem basic, with something that sounds like “Across The Universe” being generated in less time than it’d take for you to click your fingers.

In the amount of time it will take for all The Beatles to die, and then for everyone who lived at the same time as The Beatles to die, then this might not seem like a bad time. It might sound quite good. It might be indistinguishable from the original.

So too with the written word. Ask AI to create Borges’ “Library Of Babel,” with every possible combination of letters, and it might be possible—it might be already.

For fiction, the outcomes are similar, and perhaps equally horrifying. I will stress again that this is all theoretical for now, much like my robotic surgeon which won’t nick your artery, or remove the wrong kidney, or insert a toxic implant in your breast, or leave its scalpel and forceps stitched up inside you, or leave its initials soldered on your very bones for a laugh.  But let’s say you were to feed into a powerful computer every single sentence written by an author—say, JRR Tolkien.

Now, say you added into this database every other piece of surviving written work by him— letters, notes, introductions, reviews, essays, shopping lists, whatever. And then say we blended this with every single piece of Tolkien’s biographical data and its historical context. Is it possible this could be used to come up with a new piece of epic fiction as good as, and indistinguishable in style from, The Lord Of The Rings? If so, it might be a crime of sorts to deny anyone the opportunity to read that.

For some, it’s a crime to create it. But consider what’s happening right now—thanks to human agency, not machines. Take Sherlock Holmes, and the amount of literature and movies and TV shows that are still created up to the present day featuring Conan Doyle’s great detective. I read a short story by Val McDermid at Christmas, in her festive anthology. Collections of brand new short stories featuring Holmes and Watson appear just about every year. With Holmes now having passed out of copyright control in the US, you can now expect far more of these.

And for absolute clarity, if anyone offers me a chance at writing a Holmes story, I’ll take it quicker than you can light a pipe. Elementary.

Culture regurgitates. AI won’t change that.

Think about the torrents of fan fiction out there, licensed fan fiction, at that, based on any intellectual property we might think of. How many novels are out there, based on the Alien universe, alone? If Kingsley Amis did it with James Bond, then it’s fair game for anyone, really.  So, if anyone’s ever of a mind to complain about content generated based on existing properties, an appropriation of beloved characters and stories, then living, breathing humans are way ahead of AI.

To confront the thing we fear directly: can I see any instance where I might use AI? I can.

Neil Gaiman wrote a story, I think part of his Sandman universe, where someone is cursed with having too many ideas. Like most writers, I think I have this curse as well. I’ve often wondered at the word count of my ideas file, stretching back more than 20 years. It’ll probably be novel-length by now. And it grieves me to think that not all of these ideas will become finished stories—in fact, I’ll be lucky if 1% of them make it beyond the one-line concept stage to an actual page, whether that’s for people to read, or just languishing on a hard drive.

If I was facing the end of the road, and such tools were available to me, would I consider using AI to turn these ideas into finished stories, based on my own writing and style, and within set parameters? I might just be tempted to do that.

The issue is authorship, of course, and being up-front about how stories are created from now on. Publishers have a moral duty to let readers know if the tales they’ve paid for were created by a human or a machine—even if it’s genre work, pornography, fan fiction, or anything we might uncharitably call hackwork, we’ve a right to know that it’s an honest endeavor.

As a fellow writer said to me recently, it may be a matter of legality; publishers might be obliged to tell us exactly how the words got on the page. Because it wouldn’t do for them to simply skip employing and paying humans for work a machine could do… But some of them will, when the technology is ready, in maybe a decade or two. It then falls upon the publishers to be honest, to allow people to make an informed choice, and hopefully a good one.

Who knows—maybe AI will unleash the sort of technological devilry that might make the publishing industry itself redundant. If people have AI tools at their disposal, then maybe readers of the future will buy the software itself and make their own stories and novels from scratch, and not bother with books by anyone else at all. Those werewolf/regency romance mash-up stories you’ve always had in your head? Go right ahead with those, don’t bother with the middle man. Dinosaurs Vs Elvis? Complete retelling of Shakespeare as dirty limericks? Or you, as a great warrior or lover or racing driver or superhero or a Starfleet captain . . . or God? You can see how it might go. Endless reiterations and mutations.

As for the humble writer, well, I wouldn’t call myself a luddite. But AI’s rise is irrelevant to me. Even if every door to sharing my work with the public slams shut, and no-one, not even my own family, want to read it, I’ll still be writing my stories, with all their flaws, errors, tortured logic, and everything else that makes them uniquely, unmistakably mine. Even if “computer says no,” I’ll find a pen and paper.

The human impulse to create cannot be superseded by technological innovations. People still paint, people still sing, people still act, and people still write. I trust that the corresponding desire among my fellow humans to experience that art cannot be defeated either. 

This entry was posted in Uncategorized. Bookmark the permalink.

1 Response to The Bot Did It: Should We Fear the AI Revolution? (by Pat Black)

  1. rkoreto says:

    Thanks for such a thoughtful and deeply felt commentary on this topic. In addition to my fiction, I edit a series of consumer-oriented nonfiction newsletters. My CEO is already pushing me to turn the work over to AI instead of all these pesky writers! I’ll only add this: Yes, we may be getting close to developing plot and following character templates. But when I think about Joyce Carol Oates’ EQMM stories and the incredible nuances of character and setting she imposes on them, I have a hard time believing authors will be phased out completely.

Leave a comment