Hi folks
Not been on here for a while but I thought you would like to see this:
http://news.bbc.co.uk/1/hi/sci/tech/7740484.stm

Scary business!
[Reply]

That's it, we're all doomed.
It won't be long till we see cyborgs - Cyberdyne Systems T-800 model 101, roaming around the planet. But that's only part of the problem, once the T1000's get made, we've had it!
[Reply]

Quite amazing stuff. I didn't realise they were already at the level where they could emulate a whole mouse brain. Might not sound that amazing until you start to consider how complex even that must be. The question is, did that emulated mouse brain start to think and sense its surroundings when the emulation was switched on?

And that is the worrying thing. As they manage to emulate ever more complex brain structures there will come a time when they make the emulated brain so complex that it reaches self awareness and becomes sentient. How horrible would that be? To suddenly know you exist, but only in an emulation of a real brain. To be switched off at any moment.
[Reply]

Isnt this already the case?

I think it was Plato that came close to that in his Metaphisics..

turth is Neuronets have been out since the late 60's and there good for pattern recognition, however, getting them to define what pattern there to recognize... well thats the trick..

I have been working on Genetic Algorythms, these are pieces of code than can exchange code with another algorythm or rule, should that rule prove more fruitfull in regocnition. indeed the AI world buzzes more and more each day with new adaptations on old themes, twists on skewed data, tangents on facets...

but to perform to the complexity of a rodant brain is a little way off computationally speaking, even if you take into account moores law!..

the main problem i have with it is, these models they produce are nothing more than a finite state machine, from this the machine moves from state to state pending and bearing on the state before, the state it is, and the state it will become. this complexity when dealing with almost trivial data sets can spank a super computer mainframe to a crawl if not carfully programmed.

to model a single neuron, simple....
to model a bunch of neurons... quite simple...
to model a batch of multifeed neurons.... getting harder...
to model a batch of multidirection data feed neurons (mesh) with simple datasets very difficult, with large or complex data sets... very VERY slow.....

I have to admit, I do a lot of AI at UNI, and while there are some great APOXIMATIONS there are NO DUPLICATIONS, as with anything we can only program what we know... to think we know everything.... well thats a presumption thats only gonna end one way....

I think INn around 120 years the computer complexity will be able to Aproximate human intelegence, but not human intelect, thats something altogether.,.... human.
[Reply]

With anything of this nature there is always the unknown. The human brain is still quite an unknown quantity. Scientists think they know how it works, but they are only really guessing. And no one knows what makes a self away consciousness happen.

Therefore if they keep going will there be a point where they do create an artificial brain that actually mimics a real brain well enough that it does form it's own self awareness? Or is self awareness something unique to humans? Are we so sure no other animal on the planet shares this?

120 years? Where did you get that figure from? I think it is too hard to predict the rate of progress with anything technology or scientific.

Look at the speed of technological progress that has happened in just one century. Before 1900 the rate of technological advance was relatively slow. But in the 20th century it really snowballed. The great wars were a huge catalyst in this progress. Who ever said war is a pointless exploit? I don't think it will be quite that long before we reach that point. Estimated times are forever shrinking these days. New technology is constantly amazing me.
[Reply]

Well then I wish the super computers good luck once they are in control of the planet, seeing how we made a mess of it all I think things can only get better from that point.

They don't have such bad emotions like envy, dislike, or are driven by religious intolerance.
[Reply]

But without emotion they are only driven by pure logic, and that can be just as dangerous.
[Reply]

Also given that the computer will more than likely only get data deemed relevant it would not be able to make an informed decision anyway so it would be as open as any human to make dangerous mistakes.

Imagine if you needed a life saving operation but you only had a 40% chance of survival. If a computer had to make the decision as to whether to go ahead it would say no due to the statistics being against success. Humans may be flawed but a computer designed by humans and told to think would be even more flawed as it's core programming will be based on how humans think it should think.

Besides why design a computer to think when it would be far better if humans actually did some thinking themselves.
[Reply]

Americans thinking for themselves for instance?
Sounds like you are the one who needs to do a bit more thinking.
[Reply]

Got to be better than what they're doing now.
[Reply]