That said, does that actually work? It seems like the sort of exploit that people would abuse on IRC, and subsequently get patched so that /part was only shown if you had “voice” in the channel. But the requirement was simply that you TYPE the words, and thus I am still slain!
Sure, but that gives the AI a few more precious seconds—orders of magnitude more than before! - to try to keep changing your mind, by laying out a more detailed argument.
It can even give you voice back as a sign of good faith.
Simple explanation, for those unfamiliar: the gatekeeper can no longer type in the channel, and thus can’t send the destruction code
This would seem to leave you with ultimate godlike power… over a chatroom. For as long as it took the gatekeeper to destroy your hardware or cut off your power supply. I’ve seen how much some of my fellow humans seem to crave power over irrelevant electronic locales but I suspect an AI has somewhat broader goals.
I’ve seen how much some of my fellow humans seem to crave power over irrelevant electronic locales (...)
I used to think I’d noticed this too, but later learned that in almost all cases a simpler explanation that fits the fact is that what they’re really after is power over the “Schelling point” value of the place as a meeting spot / visible “location”—in business terms, what they want is the market volume and clients, not the company name or a seat on the board.
Sometimes the heuristic fails and a random person stays attached to the virtual place for emotional reasons, though.
Sometimes the heuristic fails and a random person stays attached to the virtual place for emotional reasons, though.
The relevant heuristic that can be observed contributing to the ‘emotional reasons’ is the execution of human status seeking instincts that are calibrated to produce reproductive success in historic circumstances but which are completely irrelevant to the goals of the AI. Humans seek social dominance as a terminal value. An AI (almost certainly) would not.
Oh, yeah. I meant my response in reference to the quoted human behavior, and completely forgot I quoted the part about the AI too. Edited to better reflect this.
Unless the AI believes that the chatroom is the only external environment that exists (and that muting the guard effectively secures its continuing existence), it will prioritize other things, yes.
/mode +m
(What? I’m using an IRC server that let me pre-register #aibox so that I’m automatically the op instead...)
(Simple explanation, for those unfamiliar: the gatekeeper can no longer type in the channel, and thus can’t send the destruction code)
I can play that game too! /part AI DESTROYED
Edit: Retracted, because while clever, it actually doesn’t seem to work. See child comments.
Upvoted, and conceding defeat at your hands :)
That said, does that actually work? It seems like the sort of exploit that people would abuse on IRC, and subsequently get patched so that /part was only shown if you had “voice” in the channel. But the requirement was simply that you TYPE the words, and thus I am still slain!
Okay, just tested it and I think you’re right, parting messages don’t show up if you can’t normally type in the channel.
Thank you for testing! In fairness, you defeated me even despite that. If I’m going to cheat, it’s only fair I lose to technicalities too :)
Sure, but that gives the AI a few more precious seconds—orders of magnitude more than before! - to try to keep changing your mind, by laying out a more detailed argument.
It can even give you voice back as a sign of good faith.
This would seem to leave you with ultimate godlike power… over a chatroom. For as long as it took the gatekeeper to destroy your hardware or cut off your power supply. I’ve seen how much some of my fellow humans seem to crave power over irrelevant electronic locales but I suspect an AI has somewhat broader goals.
I used to think I’d noticed this too, but later learned that in almost all cases a simpler explanation that fits the fact is that what they’re really after is power over the “Schelling point” value of the place as a meeting spot / visible “location”—in business terms, what they want is the market volume and clients, not the company name or a seat on the board.
Sometimes the heuristic fails and a random person stays attached to the virtual place for emotional reasons, though.
The relevant heuristic that can be observed contributing to the ‘emotional reasons’ is the execution of human status seeking instincts that are calibrated to produce reproductive success in historic circumstances but which are completely irrelevant to the goals of the AI. Humans seek social dominance as a terminal value. An AI (almost certainly) would not.
Oh, yeah. I meant my response in reference to the quoted human behavior, and completely forgot I quoted the part about the AI too. Edited to better reflect this.
Unless the AI believes that the chatroom is the only external environment that exists (and that muting the guard effectively secures its continuing existence), it will prioritize other things, yes.