Can you offer a situation in which consciousness is a requirement? I realise that in this discussion there is no precise definition of consciousness.
I think the behavior of something like a squirrel would probably require the same level of consciousness that a real animal squirrel possesses. There is actually a ton of complexity that needs to go into the simple animal act of being able to move around, look at stuff, and touch stuff. We haven't been able to get robots, or even simulated squirrels, to move even close to as efficiently and elegantly as a real squirrel, and that's because the approaches used thus far ( non-conscious approaches ) just don't scale.
Certainly passing the Turing test will require a level of consciousness similar to that of a human. To have a robot sitting in front of you at a table, and ask it questions about stuff, and have it respond at least at the level of a teenage human, is simply never going to happen unless the robot is genuinely conscious like a teenage human. You can accept my authority as a professional AI programmer when I say that.
I was considering AI self programming and learning from its environment and past results.
Ok, well there you go -- at a sufficient complexity level, that kind of AI will implicitly be conscious.
As soon as the things it "learns" about include "itself," it is conscious on some level. Whether or not it is conscious like a human then depends on what exactly its internal representations of itself are like, how those representations are used, and what sorts of sensory input it has access to.
This needs looking at more closely I feel. So if one assembles a system which performs a certain set of computations, it would necessarily be conscious provided the computational requirements for consciousness are met?
Yes.
Do we know what these requirements are?
We have a pretty good idea. I will make a dedicated post soon to answer this in more detail.
I am of the opinion that the earliest forms of life are conscious of their environment. To keep it simple, I am saying that consciousness began with the first single cellular organism.
There isn't anything wrong with that position, since even the first single cellular organism was vastly different than anything else in the universe ( due to how it used a sequence of computations to "live" ). Don't listen to anyone that tries to tell you cells are just like any other system of molecules, it is absolute rubbish.
Yes, this bring up two issues the first of which was pointed out by piggy. Namely that with such a broad definition of computation, a hurricane or a waterfall would be conscious. As a result of exhibiting computation of some form.
Don't worry about what piggy says, it isn't relevant. I never said all computation leads to consciousness. It is only a
sequence of computations that satisfy a very
strict set of constraints that lead to consciousness ( or rather,
are conscious ).
A hurricane can be said to perform maybe a few basic computations, since it does settle into an attractor state if the temperature and humidity are above the threshold required to keep it's thermodynamic engine going. But that's all -- a few computations. That is orders of magnitude less than what you need for even the most basic idea of consciousness, even PixyMisa's SRIP definition requires more than 10 computations ( since a transistor is roughly one computation ), he has said so himself.
And it isn't clear to me that a waterfall performs any computation at all. What are the attractor states of the waterfall? But even if it did, it wouldn't be anything near a sequence that would lead to consciousness.
Secondly, which I mentioned in the previous thread, there may be something about matter (some property) which results in consciousness when the correct combination of chemical reactions take place. Surely this property of matter would be a requirement for your position. If matter did not exhibit this property any consciousness may well be impossible.
We do not know what this property is, or distinguish a form of matter with the property from one without the property. Therefore we cannot predict that electrically mimicking those chemical reactions would result in the same effects, ie consciousness.
Well I completely agree -- the property required is the ability to settle into stable or meta-stable attractor states. It's called computation. Anything that has that property can be used to build something that is conscious. But remember, it isn't merely computation that leads to consciousness, it is a sequence of computations that satisfy a strict set of constraints.
The position that there might be something essential to the chemical makeup of neurons isn't logically consistent because there are many cases where neurons are alive yet the brain is not conscious at all. The fundamentals of consciousness must be related to the
difference between a living firing neuron in a non-conscious brain and a living firing neuron in a conscious brain. That difference, as far as anyone can tell, is merely the coordinated causal sequence of firings. Random firings lead to nothing, organized sequences of firings lead to consciousness. This strongly suggests that if you have any other medium that can support such causal sequences you have a medium that also supports consciousness.
Yes, my position is that life has manipulated matter or systems of matter in the distant past or outside our timespace system and that the property mentioned in my previous paragraph was introduced or manipulated by life in the first place.
Life does one fundamental thing that no other system does, and it is easy to understand if you think hard on it:
The computations in Life result in an increased likelihood of the system remaining in a configuration where the same class of computations can be repeated. Thus "Life" is really just a class of configurations of a system that behave in a way that statistically prolongs the existence of that class of configurations. It isn't magical ( well it is, but not like people think of magic ), it isn't special ( well, it is, but not like people think of special ), and it certainly isn't divine. In fact, it is sort of inevitable -- if you have a universe where systems with stable or meta-stable attractor states exist, and those systems can interact with each other, given enough time those systems will configure themselves to form a greater system that does what life does. It is nothing but statistics.
Life is insanely good at it, and it is really easy to prove: take an average cell and stop the processes of life within it. It decomposes into chemical mush almost instantly. Now look at yourself -- each cell in your body has been in a similar configuration for billions of years. Find me anything in the universe --
anything -- that is as fragile in its environment as a living cell yet has survived in the same configuration that long.
However, this
does take a non-trivial amount of coordinated computation. I haven't bothered to estimate the number of coordinated computations that take place in a living cell, even a prokaryotic one, but it is certainly in the thousands. Far more than what takes place in any non-living system ( excluding the systems life has created, of course, for example computers ).
PS -- to head off any irrelevant responses from the peanut gallery, I should add that a hurricane is not alive, either. The computations carried out by a hurricane lead to
nothing. It doesn't change it's behavior as a result of any of them. If the hurricane runs out of moisture, and starts to die,
that itself is the computation -- some other system could use the hurricane to determine "there isn't enough moisture here" but the hurricane itself makes no use of it. Living cells, on the other hand, change their behavior as a result of the computations that take place inside them.
