4 ms·
> I don't think AI is in any danger of taking over the world just yet. The scary thing about AI is that, even as the algorithms have greater and greater intell
by wizzwizz4 7y ago
> I don't think AI is in any danger of taking over the world just yet.
The scary thing about AI is that, even as the algorithms have greater and greater intelligence, we're still not much closer to teaching them to do what we want them to do. They can game the system better than ever, and then the universe is tiled with surgical masks.
- Kye 7y agoMaybe this is how AC finally reversed entropy. https://www.multivax.com/last_question.html https://www.multivax.com/last_question.html
- spiritplumber 7y agoInsufficient data for an answer.
- mcv 7y agoSo if AI ever takes over the world and kills us all, it will probably be because it failed to understand what we actually wanted.
- WalterSear 7y agohttps://wiki.lesswrong.com/wiki/Paperclip_maximizer https://wiki.lesswrong.com/wiki/Paperclip_maximizer https://www.decisionproblem.com/paperclips/index2.html https://www.decisionproblem.com/paperclips/index2.html
- TeMPOraL 7y agoOr perhaps because it was rotten to the core, as it was initially created for a malicious purpose (most likely advertising).
- wizzwizz4 7y agoIf "what we actually want" even comes into consideration, we did orders of magnitude better than the current industry standard. (Poor consolation, I know.) Right now, the vast majority of AIs don't even have a concept of "human desire" – probably none of them, to be honest, though some that are good at manipulating their handlers might've come close to a particularly stupid dog's understanding. This is at the core of the Friendly AI problem: https://wiki.lesswrong.com/wiki/Friendly_artificial_intelligence https://wiki.lesswrong.com/wiki/Friendly_artificial_intellig... Just because we created the AI doesn't mean it'll care about us. That's like saying a maths problem will try to make your favourite number its answer, just because you wrote it. No, you are the one who must make its answer the number you want. It won't happen by chance. Corollary: you can't patch broken FAI designs. Reinforcement learning (underlying basically all of our best AI) is known to be broken; it'll game the system. Even if they were powerful enough to understand our goals, they simply wouldn't care; they'd care less than a dolphin. https://vkrakovna.wordpress.com/2018/04/02/specification-gaming-examples-in-ai/ https://vkrakovna.wordpress.com/2018/04/02/specification-gam... And there are far too many people in academia who don't understand this, after years of writing papers on the subject.