Interview with Stephen Wolfram on AI and the Future(gigaom.com)
gigaom.com
Interview with Stephen Wolfram on AI and the Future
https://gigaom.com/2015/07/27/interview-with-stephen-wolfram-on-ai-and-the-future/
2 comments
The most interesting part of this interview is the discussion of what happens when human agency is further removed from an individual bot. If a bot has no owner, do we start viewing it as an animistic entity with rights and responsibilities? Or, do we return to a world of individual as opposed to corporate responsibility? with decentralized currency and homomorphic encryption, seems like we are pretty close to this world right now. Very exciting.
I think there will be no such thing as no owner. In worst cases, it will be the state that compensates for damages. Kind of like if you drive into a hole in a road you can seek damage compensation from municipality. There will be some kind of laws that would instantiate agency of some sort to regulate and make sure there are no wondering "entities" without owners and they would be responsible.
What I wonder is that considering that computation is universal, would it be easier to achieve human-like understanding of the environment using hierarchical models without any use of neural networks?
Can we achieve such mathematical breakthrough without making use of neuroscience or related biological fields?
Can we achieve such mathematical breakthrough without making use of neuroscience or related biological fields?
If you accept that computation is universal, then yes, you could definitely achieve human-like understanding through means that are completely different from the actual human wet-ware.
It might not be very efficient... but it's certainly possible. If the universe is isomorphic to a Turing machine, then the human brain certainly is. If that is so, then it is possible in principle to create a human-like machine with any other Turing-complete machine. It just might take you a billion years to simulate one second of human brain activity.
It might not be very efficient... but it's certainly possible. If the universe is isomorphic to a Turing machine, then the human brain certainly is. If that is so, then it is possible in principle to create a human-like machine with any other Turing-complete machine. It just might take you a billion years to simulate one second of human brain activity.
It's not obvious that either the universe or the human brain are isomorphic to a Turing machine.
There's an epistemological tendency to assume the universe is isomorphic to the most complex mechanical model available at the time.
Historically, those models seem to be reliably wrong; the universe always turns out to be more complicated and subtle than we imagine it is.
There's an epistemological tendency to assume the universe is isomorphic to the most complex mechanical model available at the time.
Historically, those models seem to be reliably wrong; the universe always turns out to be more complicated and subtle than we imagine it is.
You don't need to replicate human brain to achieve human brain-like behavior.
Well if you consider quantum mechanics to essentially be a complex probabilistic cellular automaton, then I don't think we'd even need to mirror our specific biological mechanisms of intelligence; we'd just need to run the right cellular automaton and wait. Detecting life may be more difficult though.
> wait.
"wait" sounds like find the right cellular automata.
"wait" sounds like find the right cellular automata.