Say we discover (somehow) that it's actually impossible to simulate a human-level conscious mind on inorganic substrate. What did we discover?
(e.g. we're running in a simulation, and a resolution limit prematurely halts Moore's Law; etc)
Conversation
Replying to
Yes, the question is deliberately vague as posed. Still, some fun answers in the thread. My favorite so far, I think:
Quote Tweet
Replying to @andy_matuschak
(OK, we're deep in sci-fi territory now)
Our brains don't actually do the processing needed for consciousness. They're just security tokens for getting processing power on the universal computer that simulates our world. "Organic substrate" is one of the authentication checks.
2
2
27
Quote Tweet
Replying to @andy_matuschak
Consciousness is very easy to create with inorganic computers, but will be recognized and automatically destroyed by advanced aliens/the simulation computer. Messy meat-consciousness is convoluted and escapes their detection. Organic consciousness is a hack around safeguards.
1
9
Replying to
depends what you mean by "impossible"—if we just can't figure out how to do it, for a long while people would simply say that we're missing some sort of "secret sauce," perhaps some yet undiscovered quantum phenomena...eventually though, theists would probably feel vindicated?
1
1
It's interesting to ponder what must be true of a scenario in which we don't figure it out in ten thousand years, and also of one in which we discover (positively) that it's physically impossible.
3
6
Replying to
It’s hard to imagine what that discovery would look like. It’s impossible so far, but impossible, period? How would we know?
1
1
Perhaps the same way we know that it's not possible to simultaneously specify both the position and momentum of a particle.
2
1
9
Show replies
Replying to
Hmm, maybe that the von neumann architecture is a bad fit for the highly parallel, interconnected and local computation that is needed to run a good mind. So it'd be a computational complexity issue, with the bottleneck being what can be expressed as GPU/CPU instructions.
1
6
... and I guess that by when you try build a neuromorphic silicon chip, you eventually run into engineering challenges that are best solved by a biological approach.
1
4
Show replies
Replying to
That it would be prudent to redirect resources to learning how to engineer using organic substrates!
Also, this would tell us that there is something particularly special about the properties of organic materials and we should really investigate what that might be.
6
Replying to
the word "simulate" is doing a lot of work here. simulate wrt what observables?
1
3
turing test type things say more about the communication channel with the AI than they say about the AI
2
8
Show replies