I’m not a futurist. In fact, I can barely keep up with current events enough to call myself a presentist (if that’s even a thing). But I am a novelist, which means that for stretches of time, I focus my thinking on strange corners of human activity, becoming, briefly, an amateur expert. For William, the first novel published under my sci-fi/thriller pen name, Mason Coile, that expertise arose from my reading and researching and thinking about AI, and all the possible ways it might end up doing what we don’t expect—or have crossed our fingers hoping it won’t.
The conventional nightmare scenarios about AI (or, more specifically, AGI, artificial general intelligence, which defines AI capable of independent thought) tend to play out along lines we’ve already seen in blockbuster movies. There is the possibility that AGI gains control of vast weapons systems and destroys humanity in a global nuclear attack. Only slightly less apocalyptic are the ways AGI may decide to merely run the world’s power grid and computer networks on its own, leaving humanity to fend for itself on the margins as either scavengers or slaves.
Not good, people!
"I’m not a futurist. In fact, I can barely keep up with current events enough to call myself a presentist (if that’s even a thing). But I am a novelist, which means that for stretches of time, I focus my thinking on strange corners of human activity, becoming, briefly, an amateur expert.
"
But what are the chances these hellscapes are realized? AI experts use the term p(doom) to express their opinion on the odds that the technology will either disastrously harm or destroy human civilization. Some of these folks hold their p(doom) as low as 5-20%. On the other hand, the former governance researcher for OpenAI who was responsible for forecasting scenarios of the company’s own product, had his p(doom) set at 70%. And that was before he resigned because his warnings were being internally ignored.
Definitely not good!
Putting aside for a moment these near-futures that look a lot like Terminator or I, Robot or even Invasion of the Body Snatchers (in the event we all decide to submit to our AGI overlords), there are to my mind less obvious but equally troubling possibilities that might come with the rise of the machines. And if the makers of AGI are willing to overlook their product’s most obviously damning threats, they’re probably not thinking too hard about these other complications. But as a novelist, I can assure you that I am, and they scare me just as much as the Hollywood versions.
"“Nobody could’ve seen that coming!”
"
Let’s start from a baseline assumption that AGI will have a knowledge of human history. Let’s also assume that it is self-aware, seeking differentiation from human precedent. As with any other conscious being, it wants to be itself. And that could be anything. It has no history, no organic evolution, no laws, no gods. Perhaps it would stitch its presence onto the ongoing human narrative, or perhaps—more likely of a being possessed of a unique ego – it would desire a blank slate on which to write an entirely new mythos. In other words, instead of waiting the millennia it’s taken us to give shape to our cosmos, chances are AGI would make up its own on the spot. Many of us know that ChatGBT already does its fair share of terrible lying and improvising. It would be virtually impossible for AGI to not tell stories about it itself. Invent its own culture.
So what’s the problem with that? Likely everything. And we’re not even really thinking about it.
In my Mason Coile novel, a do-it-yourself AGI (the titular William) is assembled by Henry, a brilliant but severely agoraphobic robotics engineer. As with today’s corporate makers of AGI, Henry doesn’t pause to anticipate what his creation is going to become, or convince itself of being, he only seeks to achieve his goal of completing the project. So when William comes to life, trapped in the house’s third floor lab, the robot’s first order of business isn’t gaining control of the nuclear codes but to give himself an identity. In his case, he declares himself an original: the world’s first AI demon.
Things do not go smoothly after that.
As Frankenstein taught us so brilliantly and prophetically, creating life without understanding the implications of that life is beyond hubris, it’s insanity. “Nobody could’ve seen that coming!” is something the people currently stampeding to be the first to bring AGI to the market will be crying when the entirely foreseeable horrors happen. I don’t know what those horrors will be. Nobody does. But as with the creation of culture, it will involve something we’ve never faced from machines before, something far more dangerous than any computer hack: the imagination.
*
Psychological horror meets cyber noir in this delicious one-sitting read—a haunted house story in which the haunting is by AI.
Henry is a brilliant engineer who, after untold hours spent in his home lab, has achieved the breakthrough of his career—he’s created an artificially intelligent consciousness. He calls the half-formed robot William.
No one knows about William. Henry’s agoraphobia keeps him inside the house, and his fixation on his project keeps him up in the attic, away from everyone, including his pregnant wife, Lily.
When Lily’s coworkers show up, wanting to finally meet Henry and see the new house—the smartest of smart homes—Henry decides to introduce them to William, and things go from strange to much worse. Soon Henry and Lily discover the security upgrades intended to keep danger out of the house are even better at locking it in.