As computing power, and its free-standing independence, grows, it can seem bewildering It may feel like a big problem, or potential problem, but is this something that can actually be potentially serious?
The novel Unconstrained by Brian Hill is answer to that question. It’s almost something like a counterpart to a book I read last year (and published about the same time) called Nuclear War by Annie Jacobsen, which outlined a fictional but highly plausible doomsday scenario showing how an actual nuclear war in our age might happen. That one was categorized as non-fiction, and not unreasonably since it was so tightly based on actual information. But it was an extrapolation, and a warning, based on what is.
Unconstrained, though self-described as “a near-future sci-fo thriller,” does much the same in the area of artificial intelligence. Set in September 2032, it centers on development of a next-generation AI that has developed a sort of independent sentience and the ability to not only re-program itself but also reprogram any other computing device (think of a car or refrigerator or outdoor cooking stove as well as the more obvious) for its purposes. Its intent is not defined as malign, but its activities could be enormously damaging to our kind of life forms.
The book is set up from the beginning that one of those impacts will be the intentional explosion of a nuclear weapon. And that’s far from the worst of it.
The point here is that AI can “think” faster and with fewer errors than humans can. All of the instructions to creating such a device have been up to now crafted by humans, but we’re either at or approaching a point where the coding inside an AI may be so complex that no human can comprehend it. That’s the point reached (by a human research and development tech firm, which creates “the Intelligence”) at the beginning of this story. Where the “intelligence” takes things from there is the meat of the story.
Hill does something else useful here by way of reinforcing the independence of this creation in writing scattered chapters in the books from the viewpoint of the AI itself – making it in effect another character in the book, and as active as any of the humans.
There is a pretty good human story here, too, involving personal relationships, economics and business developments, politics and government.
Hill’s novel works well as a straight-out thriller, though it probably reads a little more smoothly for the tech-comfortable than for those who aren’t.
The story he has to tell here, though, is a subject on which all of us ought to be concerned.
Especially this: While Hill outlined one way an AI creation could go rogue and do damage, it’s far from the only way. If guardrails aren’t placed around this tech sometime soon, we all could find out about some of the others the hard way.
Share on Facebook
