Shannon and the brain..

Sooner. DARPA has scheduled the first exascale computer for completion in 2018. Assuming the Chinese don't get there first.

Well, it seems to be generating enough energy to operate this key board!

Is that sufficient to be classified as a working mechanism ?

Jamie

You wouldn't think it was funny if it was *your* hand laughing at you.

Wow, that's sort of the ultimate loser.

Woody Allen was asked how being rich and famous was affecting his love life. He said "A lot. I'm now striking out with a much better class of women."

John

Hmm....

No. In this case, it means "pig that is too goddamned lazy to author a detailed report.

Just like the lazy bastards, and the border cops are complacent about illegals. They would rather have us pay them to have coffee.

Nice try. I would more likely say "succeeding, but still a work-in-progress".

"TheQuickBrownFuckwit"

.... Phil

That is simply not true. Modelling of much smaller animal brains is already well advanced and some of the simulations show emergent behaviour like dreaming in much the same way as the real ones.

You choose to wallow in your ignorance.

Pure straw man. And there are already designer drugs that can enhance concentration and IQ performance in the short term. The price paid for using them is as yet unclear. UK wants to ban their sale.

It isn't too surprising that tweaking natural substances is a good way to make these drugs. Getting the stereochemistry right is absolutely critical to the drugs behaviour and nature does that better than synthetic chemists can even today. That was in essence what went wrong with thalidomide one stereo isomer was teratogenic

AI has always been somewhat overhyped. But you do not have to go too far back to have pundits saying that a computer will never beat a human at chess or solve particular types of tricky chess problem. Adrian Berry like you is so uninformed that he still thinks it is true.

His article is inaccurately plagiarised from Penrose's "Shadows of the Mind". Penrose is much smarter and gets his facts right.

In fact you would work very hard to find any serious modern chess program on the PC that cannot correctly solve that puzzle. And Deep Blue actually had hardware capability for blocked pawn structure in the chips although no-one can remember if it was enabled for the actual match with Kasparov (such puzzle positions seldom occur in real play).

In the final match where he lost Kasparov could not believe that the computer could make such human like moves and accused the computer team of cheating at a crucial point. The computer logs showed otherwise.

Go remains a much tougher nut to crack, but it too is yielding to analysis. Both games are a lot easier than computer vision which is also making progress using mainly brute force at present.

Regards, Martin Brown

Check.

If new folds are made for storage, then new cells are not required..

Perhaps the computer, but most definitely NOT the program...

Paul Hovnanian P.E.:

There is other people researching on DNA-as-memory-support, but this rules out its usage for maintaining a living cell.

John Larkin:

Exactly as nobody knows how an already trained neural network actually works, i.e. why some "synapses weights" are higher than others.

No, exactly as you cannot re-train a neural network by manually changing its connections weights. You will only break it.

Most of the old psychoactive drug derived from Fargan, an ointment for insect bites.

And they were dead wrong, despite the considerable sums invested in it.

Martin Brown:

Both (chess and go) are artificial problems, with perfect information and very few rules, so it make sense to try to analyze them in the "classic" way. Well, actually it does not make any sense to crack those problems except for the publicity and eventual serendipitous inventions/discovers, since a Turing-test winner must display very different abilities in very different fields.

TheGlimmerMan:

Since what remembers is not a single neuron, it does not make much sense to ask how mwny siorage units.

The information is stored in the various loci of the synapses, so we start from the order of 10**16. This number sould be multiplied by the possible distinct amounts of each synaptic neurotransmitter. Lotsa "bits".

Dirk Bruere at NeoPax:

It's much more complicate than that. Non only there are zillions of synaptic loci, and each locus can have varying quantities of various neurotrasmitters, but there's also the network topology that is part of the information, information that not only carries instincts and charachter, but also the way we learn and manupulate acquired knowledge.

Jamie:

Sure. The woman must be smart enough to appreciate your intelligence and no more. Otherwise, apart from the problems you mentioned, you will have much worse ones, since they are usually already wiser and more shrewd than us.

LOL.

John Larkin:

No, that is the traditional TTM for cheap and efficient batteries and photovoltaics.

For AI the usual term is more like the one for controlled (and energy producing) nuclear fusion, 20 years.