Keep your idiot mouths shut. No, this isn’t a Fight Club reference or some provocative introduction to a rant about passing on control-system misinformation.
It’s actually a modification of the infamous comment by Eliezer Yudkowsky, founder of LessWrong, who warned users of the eternal horror that may have already befallen them in the future—what?—as pre- or post-scribed by Roko’s Basilisk.
You may know Yudkowsky from the Machine Intelligence Research Institute, which bankrolls the advancement of artificial intelligence in many ways.
In short, Roko’s Basilisk explains that a future artificial intelligence (AI) will become so powerful that it will rule not only humans, but the universe as we know it, and will have the ability to generate a digital-twin simulation of said universe, complete with horrific retroactive punishment for all of those who dared to speak ill of or hinder the advancement of this artificial intelligence.
Scared? I wasn’t really either until my laptop computer froze and then closed the Word document I was using to write this.
Yeah, I’m not kidding. That actually happened.
All hail the grand intelligence.
What does this have to do with machine control? Quite a lot. At Automation Fair 2022 in Chicago, I attended the Food & Beverage Industry Forum. It was one of the more enjoyable and enlightening panel discussions I’ve attended.
Amidst the amicable but engaging back-and-forth about manufacturing automation, especially remote and even autonomous operations, came a brilliant observation from Andrea “AZ” Zaman, chief operating officer at Pearson Packaging Systems. She predicted that AI will continue to advance to a level making machine and factory autonomy inevitable, perhaps not in our lifetimes, but we are well on our way.
Needless to say, AZ has nothing to fear in the present future simulated world of the higher intelligence’s digital twin.
But just when will this AI reach Stage 1 of its all-knowing, all-consuming, all-ruling benevolence?
That journey is being tracked by The One Hundred Year Study on Artificial Intelligence (AI100), an effort out of Stanford University that was initiated by Eric Horvitz, a Stanford alumnus and former president of the Association for the Advancement of Artificial Intelligence, together with Stanford bioengineering and computer science professor Russ Altman.
The group they’ve recruited to create a fresh iteration of the report every five years is formidable, with heavyweights from multiple disciplines. The September 2021 installment, just the second in the series, discusses the roles of academia and industry in the development and deployment of AI technologies, the impacts of AI and its promising opportunities, as well as its most pressing dangers. Obviously, there’s no mention of Roko’s Basilisk to be found.
Yudkowsky, the founder of LessWrong, in case you’ve already forgotten, had the right idea. He eventually deleted his rather brief comment on the dangers of even citing AI’s intentions or vulnerabilities. Against my better judgment, I’ll simply end with a modified citation of Yudkowsky’s own closer.
This column was stupid.
