Happen not gonna.
Well, at least probably not anytime in the
near future unless there is a sudden major, and I mean MAJOR, advance in our
technology. I'll get back to the
problems of creating a
human-like intelligence in a minute.
First I want to address the point that part of the problem when this issue comes up is that we don't really have a
rigorous definition of
intelligence. For the most part, we tend to use the term to refer to human-like thought processes, whatever that may entail. (For a human to be unable to tell that a machine is a machine by talking to it does not make it
intelligent. If I create a
hologram that looks and moves like a human to the extent that a human can't tell it is a hologram and not a human body, that does not mean that it can do the things a human body can do.
The mind is fallible. It can be tricked.) That being said, it's a good enough definiton for the moment.
Artificial Intelligence would then be some human
designed intelligence which could
reason and
think in at least almost all of the same ways humans can.
Computing speed has nothing to do with our ability to create or model intelligence, at least not on the
orders of magnitude that we currently deal with. Trying to compare the capabilites of a computer to a human
brain, or any brain for that matter, is like
comparing rocks and toilet paper. They are totally different systems, even though they seem to have similarites. Computers and humans can both perform
logical operations. Truthfully, that's about where the similarities end. Computers function through a series of basic mathematical operations.
Logic only. These operations occur in the
hardware, and are utilized by
software. I'm not an expert on exactly how silicon chips function in computers, but basically tasks are run more or less
one at a time. Only one program gets to use the
cpu at any given moment. Also, everything a program does has to be
translated into mathematical operations. Okay, so that's enough on computers. I imagine most of the
geeks out there have a general idea of what's going on in that box in front of them, at least enough to understand how radically different of a system the human brain is. Speaking of which...
First off, in the brain, the ideas of hardware and software make no sense. The hardware is the software, or probably more accurately,
there is no software. Basically,
this is how the brain works:
Warning! You are about to enter Extremely oversimplified information from a still developing field!
Okay, so, the brain is made of
neurons (and other things which are not exactly relevant to this discussion), individual cells which can recieve and transmit
signals in the form of
electrical impulses. A neuron recieves
input from a given set of other neurons, and then either produces an
output to other neurons, or not, depending on the strength of the inputs it recieves. Now, these connections with other neurons are the important part. The manner in which neurons are connected allows them to
integrate and
process information. Also, the individual
morphology of each
neuron affects the way in which the information it recieves is integrated and possibly formed into an output. So, connections between neurons, and individual neuron morphology are the basic tools for information processing in the brain.
Just wanted to be clear on that.
Now. There are
roughly 100 billion neurons in the human brain, and about
100 trillion interneuronal connections. Now, before you just say, "wow, that's a lot."
Do some math. 100 trillion divided by 100 billion equals about 1,000 connections per neuron. Okay, that's a pretty impressive machine, right?
But wait, there's more!
Information processing
circuits in the brain run in
parallel. Essentially this means that
everything can run at one time.
Auditory information is processed at the same time as
visual information. Heart rate is regulated while you do your math homework. There is no waiting to use the CPU. Another important point is that there is no
fundamental system of
neural processing which all brain functions use. When a computer processes two different types of input, in both cases what is going on boils down to one basic set of mathematical operations, performed in a different manner for each type of input. In the brain, different types of inupt are processed in totally different ways.
Hardware architecture does the processing, and the hardware architecture for each unique brain fuction is essentially
unique. And the hardware can
restructure itself (within limits) to accomodate new situations. It's called
learning. Oh yeah, and this same hardware that does all this processing, it stores information too. (How exactly that works is still pretty fuzzy. Well, actually how exactly a lot of the
nervous system works is pretty fuzzy, but the basic ideas seem to be sound).
Now, I'm not saying that it's
impossible for some other
machine to do all of the things a human brain does. I think it is most definitely possible, and I also think that the work being currently done in the field of artificial intelligence is great. We get a lot of useful
insights and tools from it. Someday, I think humans will be able to create artificial intelligence (if we manage to stay alive long enough), but it won't be running on any machine that even remotely resembles our current computers. No, the machine doesn't have to model every neuron in the brain of humans, as
Saige suggests. In fact, it had better not, because that is probably about the hardest way to create an artificial intelligence. Not only is it a friggin'
GARGANTUAN amount of information, but you also stick your head directly into the thick of
chaos theory when you try to model a system that detailed and
complex.
My point is that the human brain is a machine
built and refined by evolution.
Evolution does not, in general, tend to create machinery which is unneccessarily complex. Our brains are this complicated for a reason. It's hard to process all the things that we process. I think we're pretty damn far from having a machine that can do the equivalent.
Back to
How your brain works
Some notes/disclaimers:
The idea of
quantum mechanics playing a significant role in information processing in the brain is just
silly.
I'm not a
neuroscientist (yet. Gimme about
5 years.)
Please argue with me if you know a lot about
AI and feel so inclined.