Re: Something more interesting, please!

Tim Tyler <seemysig@xxxxxxxxxxxxxx> wrote:
Alpha wrote:

Therefore, the two classes of phenomena, neurons and phenomenal
experience are not *identical*; to assume or posit such as Curt fdoes
is a category error.

Uh, the idea is that the configurations of neurons *produce* the
phenomenal experience - not that they are the same thing.

No, the idea is that they are the same thing, not that the configuration of
the neurons *produce* something. This is the whole complexity of the topic
and why you see it debated ad nauseam here (and elsewhere in philosophy).
It's extremely counter intuitive to understand that they must be the same
thing. If you believe them not to the same thing, it creates a whole
string of paradoxes which falls out as the hard problem of consciousness.
However, if you grasp the idea that there's nothing preventing them from
being the same thing other than our intuition, then you find the solution
to the hard problem of consciousness. Once you realize that they are the
same thing, then there are no contradictions or paradoxes and there is no
hard problem of consciousness. Alpha is a prime example of someone who
can't get past his intuition on this topic. Most people it seems can't
past their intuition on this topic. Many people in AI can't get past their
intuition on this topic. But it doesn't prevent them from doing good work
on AI because they just believe they are building zombies and figure one
day someone else will figure out the hard problem of conscious.

My continued frustration is that there are all these people waiting for the
hard problem of conscious to be solved when it's already been solved and
the only problem is that no one has been able to explain it to them in a
way they can understand it. Dennet wrote a huge book on it, and even with
his education, and respect in the field, people can read his book and still
not understand it. I don't think anyone has gained any understanding on
this topic by reading my posts. Either they understood it before they read
my views on the subject, or they don't understand it and my posts don't
change anything about what they believe.

I've noticed that a lot of comp.ai.philosophy fare is on
about this level :-(

Yeah, it seems to be the only topic that keeps people talking. :)

How about some posts on something interesting:

e.g. will the first AI arise out of a proofreading project, a customer
service shop, a search assistant, a robot project, a NPC in a
game, out of something to do with pornography, or in some other way?

I vote for a government funded robotics project for the military.

or: we have a whole WWW of textual training data - how best to use
that to make an AI that does more than proofreading?

You can't understand English to the level needed if you can't ground the
meaning of the ontology to our existence in 3D space/time. Words,
pictures, and videos from the Internet are not enough for an AI to
understand what it all means to the level needed to do constructive
proofreading. This is why machine language translation has never happened
beyond a very crude level. AI can produce better spell checkers and
grammar checkers, but it runs out of steam when it can't understand the
content of anything it's reading because it has no grounding. The
information it needs to understand all the text on the internet is not
actually on the internet, it only exists in the heads of the humans who
built the internet.

The only way to get that grounding is to build a learning machine and give
it sensors and effectors connected to the real world (eyes and ears, arms
and legs). Projects which have attempted to give a machine the grounding
by hand-coding the information into the machine have gone no where
interesting. I believe the information needed is far beyond what a human
can understand. What we understand, and talk about, is only the tip of a
much larger iceberg which is beyond our understanding. That is, though we
can understand the concept of how it works, all the specific details are
beyond our ability to hold in our head. We can't possibly know enough
about all the information in our head to hand-wire all the neurons together
correctly in an AI so that it has the understanding it needs to match human
undersanding.

If a human were to sit down, and try to do a brain dump by witting down
everything they know (every little detail about everything they can
remember), even if they had help and was prompted to help them remember
everything they knew, the resulting document would still only be a tiny
fraction of all the information that existed in their head. It's all that
hidden information, what people might talk about as their subconscious,
that is beyond our ability to self-understand. It's all those subtle
interconnections between concepts that cause one concept we think about to
trigger the next one that's way beyond our ability to understand about
ourselves. But without all those subtle interconnections guiding all our
behavior and all our thought processes, we wouldn't have the understanding
needed to do the proofreading.

The only way to get all those subtle interconnections defined correctly is
to build a learning machine, and the only way to get human level
understanding is to ground it all to the reality we live in, which is our
interaction with 3D space time, full of effects (like gravity, and mass,
and light, other humans, cars, rain, clouds, birds, etc). So you have to
build a robot of some type, and the closer that robot is to being
human-like, the more its understanding will match the understanding of
humans, and the better it will be at tasks like proofreading and
translation.

--
Curt Welch http://CurtWelch.Com/
curt@xxxxxxxx http://NewsReader.Com/
.