> “Philosophers care too much about deep-sounding meta-questions, instead of sticking to what can be observed and calculated.”
Philosophers are so lost in their consciousness/mind/body debates they can't appreciate the simple framework of reinforcement learning. RL would require simpler initial premises - no need to define consciousness or mind vs matter - just agent, playing in the environment, maximising rewards. All measurable and testable in simulation. Using RL AI's even bested humans at go, learned to walk and play complex computer games. But more importantly, RL covers all the aspects of mind vs body and consciousness. RL has more precise terms and they would benefit a lot from understanding this framework.
Among the philosophers of today I mostly align with Daniel Dennett and Dawkins. So there are still a few with the right mindset.
We can simplify the problem consciousness/mind/body simulation much more than what you think if we use your approach.
Give me two programs that produce "seeing red light" 200 ms in a loop and "hearing 400 Hz pure sine wave" for 200 ms from the point of view of person who is not thinking anything particular, accessing his/her memories or feeling his body just at the moment.
Since all computers with limited memory can be reduced to finite-state automaton, we don't even need complex models of computation like Turing machine. Just bunch of states and state transitions is enough.
There is no need for machine learning or the ability to solve problems. Just simple program that is being fully self aware and conscious experiencing simple sensation for a very short time in a infinite loop is enough. Because we can fully control the input, we can eliminate all states and state transitions not visited during the loop.
> “Philosophers care too much about deep-sounding meta-questions, instead of sticking to what can be observed and calculated.”
Philosophers are so lost in their consciousness/mind/body debates they can't appreciate the simple framework of reinforcement learning. RL would require simpler initial premises - no need to define consciousness or mind vs matter - just agent, playing in the environment, maximising rewards. All measurable and testable in simulation. Using RL AI's even bested humans at go, learned to walk and play complex computer games. But more importantly, RL covers all the aspects of mind vs body and consciousness. RL has more precise terms and they would benefit a lot from understanding this framework.
Among the philosophers of today I mostly align with Daniel Dennett and Dawkins. So there are still a few with the right mindset.