Date: September 14th, 2026 12:23 AM Author: Exciting Bawdyhouse
It's Bargaining. They already know superintelligence is here, that AI is solving problems collective humanity couldn't for centuries. They can sense what is coming next. The "slow down" discussion allows them feel like they had a choice, that they sat down in solemn council, weighed the risks, and attempted to negotiate with the future. It's completely predictable COPE.
Date: September 14th, 2026 12:35 AM Author: aphrodisiac forum jew
How does your understanding of the operation of LLMs lead to a "10% chance" that "AI" causes the extinction of humans, or suddenly become "super-intelligent"?
Date: September 14th, 2026 12:50 AM Author: vermilion fragrant pervert
You should reconsider your position given, at a minimum, everything that is coming out from people actually in the rooms building these things combined with the information that has trickled out the past couple months regarding the hugging face attack and the dsewiki attack. Have you read Yudkowsky’s book?
Date: September 14th, 2026 1:10 AM Author: Exciting Bawdyhouse
I don't think the idea that AI is likely to cause full human extinction follows from any of the things these people are talking about. It is too big of a leap, and requires extrapolation. Most of what they are saying is that the unreleased unguardrailed systems during testing can be vastly more capable and much more deceptive and harder to control than outsiders realize. I believe that, but I still need an argument for why that terminates in everyone dying.
Date: September 14th, 2026 1:17 AM Author: vermilion fragrant pervert
Please read Yudkowsky’s book and then let’s talk again. It’s an easy read and should take you a week, maybe two if you go into the technical appendices.
And we are talking risk not certainty. From your own reply you should assign a 10% likelihood.
Date: September 14th, 2026 1:29 AM Author: Exciting Bawdyhouse
I don't think it would convince me, I think your number is too high. I would put it at 1-3%. I have seen these Yudowsky-style arguments before, they require too many assumptions on things where there is too much uncertainty. But I might read it anyway, I think it would still be interesting.
Date: September 14th, 2026 12:37 AM Author: Exciting Bawdyhouse
Are you saying I said it would lead to the extinction of humans? Because I didn't.
If you are trying to say it isn't "super" yet, then what is your definition of superintelligent? I would say solving problems humans were too low iq to solve makes it "super".
Date: September 14th, 2026 6:32 PM Author: Exciting Bawdyhouse
A device that is engineered and designed to do basic computations through known simple algorithms is not superintelligence. AI is different because it generalizes.
Date: September 14th, 2026 1:50 AM Author: Saffron roommate sound barrier
A couple months ago you didn't know anything about AI, now you're at the stage where "it's just predicting what the next word is gonna be" lol you're hopelessly behind on this stuff
Date: September 14th, 2026 7:04 PM Author: 180 principal's office
close but the grief is actually over the fact that they're not competitive against open source and there are literally trillions of dollars about to go poof.