Just wondering with all the news and stuff about supercomputers being as or more powerful than the brain. How do they know? And what factors are used? Surely, a computer doesn’t have emotion… does that make us faster? Etc
A neuron can fire as many as 100 times per second (Hz). The signals move at ~30 meters per second. This is considerably slower than the 5x10[sup]9[/sup] Hz and 210,000 km/sec of a speedy processor. Parallelism, clever wiring, and highly evolved algorithms do a lot to speed up what would otherwise be a unacceptably slow system. Unlike modern computer architecture, a single neuron “processor” may connect with over 1000 other similar neurons.
Mhz as it refers to computer chips has to do with clock cycles- each clock cycle some work can be done. How much work is done during that clock cycle is dependant on the processor.
i.e. a AMD 2100+ is something like 1.8 GHz (I forget exactly) and a 2 GHz P4 is 2 Ghz. The AMD Chip is faster in terms of computing power while the P4 has a faster clock speed.
Also at one time there was a P133 Mhz chip and a AMD 133 k4? (486) chip. The clock speed was the same in both but the P133 was a generation ahead and was faster (again in terms of computing power)
So clock speed alone doesn’t mean much unless you are compairing identical chips. What you need is something like instructions per second for a fair comparason.
Now take this to the human brain - which is based on a totally different arcutecture and you run into even more problems like AFAIK there is no clock cycle - things just go pretty much continously whenever they feel like it.
You can’t meaure the brain in clockspeed and hope to gauge it’s power from there. It won’t work.
The brain is a massively parallel neural network with asynchronous processing. It is exceptional at doing some things such as pattern recognition (sound, image, etc.). The human brain is the only thing in the world that can originate new ideas.
It’s kind of like asking how many horsepower your computer’s CPU runs at. We’d get the implication of what you were asking about, but the measure doesn’t really apply because the mechanism by which it gets work done is so different.
Just asking after reading this article.
On an interesting sidenote, many attempts are currently underway to fully model human cognition with a good level of success. These “cognitive architectures” act just like one was modelling program behaviour on a modeled computer architecture. People generate models of certain activity, check them against empirical results and assess the model. Check it out:
ACT-R is probably teh most succeful effort: http://act-r.psy.cmu.edu/
I was fooling around with neural networks (a very crude brain simulator) a while back. What i did was simulating these networks on my 2GHz computer.
I was able to simulate a “brain” consisting of a couple thousand neurons (brain cells) realtime (ie 100 updates per neuron per second). For comparison a bee (the flying, honey making kind) has somewhere around 50000 neurons, IIRC.
Parallelism is a powerful tool.
Thought it might help put things in perspective.
(I know this isn’t exactly good science but i found it interesting at the time)
Computers are big number crunchers. It’s like a machine with a big crank. You put numbers in, turn the crank, and numbers come out. The only thing about computers is that the crank turns really really fast.
The brain works on a completely different mechanism. Rather than just doing one thing and doing it well, the brain has millions and millions of interconnections, so many in fact that we can’t understand how the thing works.
The big chess battles are a good example of how the human mind and the computer think differently. The human mind developes broad strategies, and looks at the board to get a mental “picture” of it. The computer on the other hand can’t picture anything, and just does a brute force bizillion calculations to try and figure out its next best move. Because a computer can do so many calculations, it can beat the pants off of someone like me in chess. But for someone who is a real chess master, the computers are said to be very much lacking in strategy. They don’t really formulate a plan. They just calculate the heck out of everything.
So, if you want to figure out flow equations around the space shuttle get yourself a computer. But, if you want pattern recognition, there is still no better pattern analyzer on the planet than the human brain.
A pentium can add up a few hundred million numbers per second. Your brain can’t even come close to that. However, put an apple into your right hand and your brain instantly recognizes it as an apple, out of the millions of arbitrary possible objects that it could be. Hook a computer up to a camera and have it try and do the same thing and you’ll find out just how miserable machines are by comparison.
Urban Ranger, I disagree with your statement that
There are aparently programs that can write poems, and robots that can paint in various techniques (Including abstract) from a video camera input.
I wish I had an online cite, but I can point to a couple offline cites. The first is (I Think) “The Age of Spiritual Machines”, By Ray Kurzweil; and “The Fabric of Reality”, by David Deutsch. I think both mention that and give all the proper cites.
And who wrote these programs, pray tell? A computer which spontaneously came into existence?
Artificial neural networks contain electronic “neurons” that think pretty much the way a real brain thinks, to some degree. The interesting thing about these is that once the nueral network gets big enough, it starts making wave form patterns very similar to what you see in the human brain (and animal brains). Unfortunately, by the time they get to this level the interconnections have gotten so complex that we fail to understand exactly how they work.
It’s all completely artificial, and probably dumber than a fly’s brain, but these machines do mimic the thought processes that go on inside the human mind. It’s only a question of degree. They aren’t quite complicated enough to create rap music yet. But they do think. Sort of.
Are you familiar with Searle’s Chinese box?
The poem, or anything really, doesn’t become an idea until someone reads it. All the program does is generate numbers according to rules set out by a human; it’s not reasoning, it’s not thinking.
Yes, I remember those AMD chips! The early ones were designated 5x86-133 and later they called it the 5x86-P75, to give you some idea of its equivalent in the then-new Intel Pentiums. I sold a boatload of those puppies on eBay, and built at least a half-dozen systems around the chip. The tough part was finding a mainboard that would support the 4X clock multiplier the AMD needed to push its 33-MHz external speed to 133-MHz. For sure, it was the hottest 486 you could buy!
(Sorry for the minor hijack…)
Those programs merely apply the creative ideas that a human programmer gave them. There’s nothing “original” in what the computer does.
You’ll also find that with most of those programs, you have to weed through dozens of unartistic outputs before finding one that seems artistic (the poems are often laughably bad or even unrecognizable as poetry). Keep in mind the million monkeys.
Urban Ranger wrote:
“No man has ever had an original thought.” — E. A. Poe
Human “creativity” is just taking pieces of knowledge (database) and constructing it in different patterns. It is complex, and probably impossible currently to predict this in any fashion, but creativity by implication should be possible by advanced enough machines.
If a human being did have an entirely original thought, it would be incomprehensible to everyone else because they would lack a familiar frame of reference, an existing departure point.
However, human beings have ideas that are 0.5% or 1.2% original all the time, and it would be a boring world with no progress if they didn’t.*
It’s more than mixing the same elements in random new ways.
- and yes, I pulled those percent figures out of my ass, but you know what I mean.
But how many “Gigs” does the average brain have?
No, not really, but I would not want to hijack this thread further by a debate on orignial thought.
I don’t see figures pullled “out your ass” as being evidence for this “something from nothing” discussion, perhaps you would be willing to create another thread on this?