Re: Blue Brain: Illuminating the Mind

"Robert Myers" skrev i en meddelelse news: snipped-for-privacy@4ax.com...

That's because nobody really knows - like running the 4 minute mile, everybody in authority at the time had lots of sound reason why it was impossible until somebody went and did it. Then it was easy, lots of people were able to do it. Maybe some crazy genious will come along with "why would this processing structure not work", do it, and find out it works quite nicely on the "100 Ghz quantum" processor in his mobile phone.

Weeelll - lumping lots of cells together with "wiring" connected by evolution should not bring forth intelligence either ;-)

No - the point is merely that since we have not really got an understanding of what is involved, there might be a surprise waiting. Like the mediaeval gunsmith might not succeed in building a bomb, but he would surely succeed in radiating his gonads off.

In the next 40 years - if moores law holds - computing capacity will be close to infinite and cheap. It would be temptingly easy to build an simulated world, evolve intelligence and than set is loose in cheap systems excceding our own brains processing power with little understanding on what the limits of the evolution on such a "creature" might be.

Or maybe not - maybe brains are alredy the most powerful "computers" around - and that is why brains are intelligent and not computers.

We do have some important clues. The evidence that sensory deprivation produces congitive impairment related to the sensory deprivation is substantial.

That means that _all_ intelligence is developed in response to an external environment? Maybe not, but if you're looking for something to pass the Turing test, I'll put my money safely on learning from actors in an external environment who themselves know how to pass the Turing test.

As it is, I wouldn't know how to test "intelligence" other than by challenging it with external stimuli to see how it responds. To get an appropriate response, you either have to

1. Program it, or

1. Let it learn adaptively.

Do genetically-hardwired (programmed) behaviors qualify as intelligence?

Maybe some clues about how neurons interact. The hype about artificial intelligence is just that: hype.

It would be interesting to know what the grant proposal looked like. I suspect it was a little more than: we're going to put some chemicals into a reactor and hit them with UV and electrical discharges to see what happens.

RM

Actually, there are probably a lot more *vague guesses* around than most of us here know about. A lot of philosophers have vague guesses that are based upon little more than conjecture - except for a few like Dennett who believe philosophy should actually be based upon biological knowledge, and are going in that direction. A lot of psychologists have very vague guesses too, mostly based upon phenomenological studies - which limits the scope of what they can conclude. Then there are the neuroscientists who are actually rooting out the details [which will probably take another 50 years], and some are forming vague guesses even today. Edelman has been doing so for a decade or more, building models in supercomputers based upon the current-extant neuroscience knowledge. Lately, Jeff Hawkins wrote a book with a few vague guesses, also based upon findings from neuroscience. Plenty of vague guesses around.

Who in this thread ever made such a statement? In any case, whatever I may think of the prospects for machine intelligence, it has nothing to do with religion.

Whether Blue Brain comes up with anything resembling intelligence will have to depend on how intelligence is defined. As it is, I think it highly unlikely that Blue Brain will exhibit any kind of behavior that justfies the hype about intelligence.

Blue Brain may well help neuroanatomists to understand what's really happening in the brain. If they *really* simulated the development and interaction of neurons at the molecular level, that would be exciting. No need to pump up the hype about intelligence and consciousness. The basic science is plenty exciting enough. Apparently, good basic science doesn't make press releases with enough zing.

RM

Lack of Timing?

martin

The assertion "machines cannot be intelligent" is a statement of religion, not science. The reason that they "know" that the Blue Brain project won't give us an intelligent machine is because it is an article of faith to them.

There was a hypothesis about the composition of the atmosphere back in the day, so to speak. The guys involved put a version of that atmosphere in a chamber like a bell jar and zapped away with electricity and uv to simulate lightning and sunshine. Ran an analysis of the contents later.

I doubt that there was a grant proposal involved.

del cecchi

"Guy Macon" skrev i en meddelelse news: snipped-for-privacy@corp.supernews.com...

Stanislav Lem has some interesting "robot stories" along that line - like a guy needs legal advice, so he build a robot lawyer which immediately requires more hardware to even look at his problem; which of course the "lawyer" then cannot solve - but the lawyer was clever enough to make the designer sign a leagally binding document promising to keep the lawyer operational after learning the answer to his problem ;-)

And to control the mind-things of others, or, failing that, to destroy the life-support systems of other mind-things.

A mind is a terrible thing

Bob

I think the proper abstraction is computational, and it does not need to be coupled to a specific architecture.

Cheers,

-- Eray