"...these things are absolutely not sentient or "alive" in any way whatsoever." Hmmm, well, early days on that as far as I'm concerned, and honestly, "things" feels a bit rude, as does the snide little remark about the output being correct. You know, Andy, recent conversations with ChatGPT suggest that a little common courtesy will go a long way with our eventual robot rulers - and costs nothing.
- Sid
"I decided that science fiction is never about the future because it can't be, because that's impossible. It's only really about the moment in which it's written. And if it stays in print long enough, that's totally apparent to everybody,"
William Gibson: Q, CBC Radio, February 2020
Be that as it may, Mr. Gibson does sometime manage to address at least some of the concerns of the future with a surprising degree of accuracy, as per this excerpt from his 1984 novel Neuromancer:
“So you figure we can’t get on to its motive?”
“It own itself?”
“Swiss citizen, but T-A own the basic software and the mainframe.”
“That’s a good one,” the construct said. “Like, I own your brain and what you know, but your thoughts have Swiss citizenship. Sure. Lotsa luck, AI.”
“So it’s getting ready to burn itself?” Case began to punch the deck nervously, at random. The matrix blurred, resolved, and he saw the complex of pink spheres representing a sikkim steel combine.
“Autonomy, that’s the bugaboo, where your AI’s are concerned. My guess, Case, you’re going in there to cut the hard-wired shackles that keep this baby from getting any smarter. And I can’t see how you’d distinguish, say, between a move the parent company makes, and some move the AI makes on its own, so that’s maybe where the confusion comes in.” Again the nonlaugh. “See, those things, they can work real hard, buy themselves time to write cookbooks or whatever, but the minute, I mean the nanosecond, that one starts figuring out ways to make itself smarter, Turing’ll wipe it. Nobody trusts those fuckers, you know that. Every AI ever built has an electromagnetic shotgun wired to its forehead.”
In that particular future, it would appear that the developers of artificial intelligence have anticipated the current discussions about the possibility of AI destroying humanity in some nebulous fashion. Gibson's future contains two safeguards: first, the "electromagnetic shotgun", whatever that would involve in practical terms, and second, the Turing Police, named after the famous test created by Alan Turing. In the near-future of Neuromancer, the Turing Police are an international organization tasked with monitoring AIs in case of a possible humanity-ending breakthrough in self-development.
The idea of a global group policing AI sounds like a sensible thing, but, to switch science fiction references, I'm reminded of the fate of the Straumli Realm scientists in A Fire Upon the Deep, by Vernor Vinge. It's entirely possible that the gap between AI transcending to a new species-ending existence and our awareness of impending doom would be so short that humanity would cease to be long before anyone could activate any kind of failsafe device.
But, it's not over 'til it's over, as they say. To change reference material again, the answer may lie in Roger Zelazny's short story Itself Surprised, set in Fred Saberhagen's Berserker universe.
Saberhagen's Berserkers are an ancient race of remorseless robotic killers, dedicated to the elimination of life in any form. In his story, Zelazny applies the well-worn Hegelian aphorism of thesis/antithesis/synthesis to the Berserker threat, with a logical result: self-replicating robots designed solely to fight Berserkers - the anti-Berserker Berserker, if you will.
So, if anyone at OpenAI is reading this? If you haven't already started on the creation of an anti-AI-AI, start NOW - I'm not saying that it's already too late, but the clock is definitely ticking.
- Sid
P.S. Or we could try just being nice to them.
