4 ms·

> And at that point, I would not characterize this effort to destroy humanity as one purely “AI”-created. It would require enlisting a multitude of people desir

by kypro 4y ago

> And at that point, I would not characterize this effort to destroy humanity as one purely “AI”-created. It would require enlisting a multitude of people desiring this “omnicide”, and human hands will be bloody as well. Why would a super intelligence require humans?

corobo 4y ago
This is the real question. We should be seeding it with the idea that humans are cute. Any plea towards or away from extermination will end with extermination. We'll either look like an easy target or we'll look dangerous and need taking out. If we're cute af in the eyes of AI we'll be kept as pets.
kibwen 4y ago
Because an omniscient brain in a tin can is still a brain in a tin can. It doesn't matter how intelligent something is if it can't otherwise affect the outside world. It's like executing a pure function: regardless of what it's computing, the computation is pointless because it cannot be observed. Don't worry about AI. Worry about humans with access to AI.
- kypro 4y ago
  I think you're probably right, but this should be treated as an unknown, and wouldn't be something I'd be comfortable to assume. You could imagine a fish arguing that if something can't swim fast and doesn't have sharp teeth then it poses no threat regardless of it's intelligence. But this is just because fish evolved to not be eaten by bigger fish and its intelligence and limited knowledge of the world makes it unaware of the other threats they could face. So while I agree it would be hard to imagine, a super intelligence in theory might find a way to affect the outside world. Perhaps this could be through some unknown physics, or more conceivably, perhaps it would first pose as a helpful AI showing humans how to build advanced robotics only to later co-opt those robotics for its own objectives.

4 ms·

> And at that point, I would not characterize this effort to destroy humanity as one purely “AI”-created. It would require enlisting a multitude of people desir

by kypro 4y ago

corobo 4y ago
This is the real question. We should be seeding it with the idea that humans are cute. Any plea towards or away from extermination will end with extermination. We'll either look like an easy target or we'll look dangerous and need taking out. If we're cute af in the eyes of AI we'll be kept as pets.
kibwen 4y ago
Because an omniscient brain in a tin can is still a brain in a tin can. It doesn't matter how intelligent something is if it can't otherwise affect the outside world. It's like executing a pure function: regardless of what it's computing, the computation is pointless because it cannot be observed. Don't worry about AI. Worry about humans with access to AI.
- kypro 4y ago
  I think you're probably right, but this should be treated as an unknown, and wouldn't be something I'd be comfortable to assume. You could imagine a fish arguing that if something can't swim fast and doesn't have sharp teeth then it poses no threat regardless of it's intelligence. But this is just because fish evolved to not be eaten by bigger fish and its intelligence and limited knowledge of the world makes it unaware of the other threats they could face. So while I agree it would be hard to imagine, a super intelligence in theory might find a way to affect the outside world. Perhaps this could be through some unknown physics, or more conceivably, perhaps it would first pose as a helpful AI showing humans how to build advanced robotics only to later co-opt those robotics for its own objectives.