3 ms·
> Our theory says that if we want to decide whether or not a machine is conscious, we shouldn't look at the behavior of the machine, but at the actual substrate
by visarga 7y ago
> Our theory says that if we want to decide whether or not a machine is conscious, we shouldn't look at the behavior of the machine, but at the actual substrate that has causal power. For present-day AI systems, that means we have to look at the level of the computer chip. Standard chips use the Von Neumann architecture, in which one transistor typically receives input from a couple of other transistors and projects also only to a couple of others. This is radically different from the causal mechanism in the brain, which is vastly more complex.
This riles me up for so many reasons. Transistors (and computers) are capable of emulating more complex systems. Looking at a transistor you will not understand why your mail can'd be delivered, nor will you understand consciousness from an individual neuron. Consciousness is related to behaviour, and both are related to survival and the environment. AI's don't have such an environment as of yet so they can't be conscious yet, but they could be. There is no magic dust in the brain.
> As long as a computer behaves like a human, what does it matter whether or not it is conscious?
If it walks like a duck, ... But seriously, why should consciousness be so special as to require a quality that can't be observed through behaviour (understanding that it is non-physical, non-observable)? P-zombies are just a thought experiment, and a bad one. Why should consciousness exist? To protect the body, to self reproduce, to exist. It exists to exist and evolve. And this is done by behaviour, by acting in a smart way. How would p-zombies come to be, if not through fighting for survival? Without the evolutionary mechanism there is no explanation for consciousness, and the evolutionary argument is sufficient. Consciousness is the inner optimisation loop, the outer one being evolution. They both work for survival, for their own sake.
An AI could repeat the same process by evolving as a population of agents cooperating and competing between themselves. They would have to be inside an environment that is complex enough, and they should be subject to evolution. It will be a consciousness, just not a human consciousness, which is tied to our environment, which is made in large part of other humans.