Did You Know?

[mcleano]

Was made to watch this is school but had some interesting facts... particuarly at 5:23 ...

" Predictions are that by 2023 a supercomputer will be built that exceeds the capability of the human brain. "

Don't particuarly understand that one myself.

Anywho enjoy if you want... http://www.youtube.com/watch?v=ljbI-363A2Q

[helios]

"We're living in exponential times" is inaccurate. Actually, the entirety of human history is an exponential curve; it's just that only recently we're reaching the interesting parts.

I don't buy the "half of the technical knowledge you learn in the first of a four year period will be outdated by the end of the fourth". New information doesn't necessarily outdate old information, and there's plenty of things that aren't that dynamic. Not to mention how long it takes for some new technologies to be embraced. For example, many algorithms in widespread use today were published more than twenty years ago.

As for the supercomputer thing, I don't think so. There's still things the brain does very efficiently that we have no idea how to do. This is a problem in the algorithms, not in the hardware.
Okay, so you have basically a programmable human brain. Can it parse natural language? Can it parse any sensible graphical input? No? Then what's so special about it? We already have computers that are faster than the human brain at certain tasks. Making one that's just faster isn't all that impressive.
And how did they measure the flops of a human brain, anyway? I don't suppose they told some guy "do 1.0+2.0 a million times".

[chrisname]

There's no way a computer could ever be more powerful than a human brain. That simply won't work. Computers do maths (hurr, they compute). Brains do... chemical reactions.

When computers are capable of original thought, emotion and developing communication between themselves (that is, evolving a complete language. Binary and networking are things we as humans came up with, so they don't count).

[jsmith]

http://www.networkworld.com/news/2009/111809-ibm-brain-simulations.html

[chrisname]

A cat brain is different to a human brain. Cat's aren't sentient (that is, they don't know of their own existance). Cat's don't have complex emotions.

[Duthomhas]

Bull. Animals are sentient.

BTW, your definition needs some help: http://en.wikipedia.org/wiki/Sentience
So few people today (at least in first world countries) spend any significant quality time with animals -- these same people then have the cultural hauteur to proclaim that animals are soulless or incapable of complex thought or emotion. People who live among animals know better than that blather.

Also, I disagree with IBM's blithe assessment that their computer "exceed[s] the scale of a cat's cortex". In what way does it exceed a cat's cortex? If they have any scientific data supporting such a statement, it necessarily must be pretty narrow. Computers are so flippin' stupid that (concurrent) processing power and the ability to collect information is the only thing they can tout.

I do think that computers will become much more powerful than we imagine now... and that they can even pass themselves off as living beings, but that's all programming.

Heck, there's even an annual contest to do that now...

Gotta take the kids to the school bus.

[helios]

chrisname confused "sentience" with "self-awareness". Animals may be sentient, but most aren't aware of themselves or their own awareness. This is demonstrated using the mirror test.

In what way does it exceed a cat's cortex?

I suppose this is all measured in bogons per second.

Computers are so flippin' stupid that (concurrent) processing power and the ability to collect information is the only thing they can tout.

Well, brains, being neural networks of massive proportions, are exceptional at parallel processing.

This whole thing about comparing brains to computers is one of the most common mistakes laymen make when it comes to computers. It's not like comparing muscle to hydraulics (although [I may be wrong] muscle may win in weight-to-force ratio). They're completely different classes of machines

[Duthomhas]

Exactly.

http://saralovesdogs.blogspot.com/2007/08/id-rather-talk-to-computer.html
http://en.wikipedia.org/wiki/Turing_test

:-)

[helios]

http://www.youtube.com/watch?v=KyLqUf4cdwc

[chrisname]

This whole thing about comparing brains to computers is one of the most common mistakes laymen make when it comes to computers

Exactly! They simply are not the same thing. Computers take input and make output. Brains are totally different.

And yes, I confused sentience with self-awareness. And no, animals aren't self-aware. Humans are the only self-aware animals alive right now (that we know exists; not to rule out aliens or anything because I strongly believe they exist) AFAIK.

Also, you can argue that animals do or don't have "souls", but I don't believe in the concept of a "soul".

Finally; animals don't have complex thoughts. They don't think about the past (unless something "triggers" that thought, e.g. a dog seeing a dog that it fought before) and they can't think about the future. All animals tend to be concerned with is the present, e.g. "I'm hungry." AFAIK that's about it, at least, for dogs. Anyone who had spent any amount of time around at least one dog should realise that they don't plan anything. Animals run on instinct alone.

Here's something for you. Essentially, everything that lives, lives to make more of itself. I mentioned, to a friend, a type of fly that lives for only a few hours -- just long enough to reproduce. It doesn't even eat or drink. It's born, it lays eggs, then dies. His response was "What is the point of that?"

What is the point of anything? All you do in life leads up to one thing - IRL fork().

Also, I think some apes are almost self-aware. I watched an excellent program about apes and humans and why we're smarter than they are (apparently, and I guess it makes sense, because our jaw muscles are attached to our skull around the temples, as opposed to at the top of the head. IIRC this allowed the skull to become larger. Evolution is cool, eh?)

Oh and do spend a lot of time around animals; I own a dog and used to own a rabbit (it died).

Also, you have kids?

Edit:
Animals are capable of emotion. I know that. Hell, my dog even makes quite clear facial expressions of curiosity, joy or miserableness (usually when I won't give him something). What I said was "complex" emotion such as what humans refer to as "love." They don't have complex thought in that... well, put it this way. Have you ever seen a dog pondering it's own existence? Do dogs have civilisation? No. Dogs have packs. Everything they do (and I respect and agree with this immensely) is for the survival of their pack. Not even for themselves. They even drive away (most people think they kill, but they rarely kill each other) weaker pack members; which most humans wouldn't agree with.

[Duthomhas]

Ah, I shouldn't have goaded this discussion. (Sorry.)

I can't debate with amateur philosophers.
The production of new scattered thoughts gets in the way of learning.
Bye.

[helios]

Exactly! They simply are not the same thing. Computers take input and make output. Brains are totally different.

Actually, in that regard brains are exactly like computers. The difference is in how they process the data, and I'm not just talking about differences in chemical and electrical processes. Computers use chiefly linear processes, with very few parts parallelized, while brains are almost fully parallel, if that statement makes any sense.
And that's just starting. There are even more fundamental differences. For example, brains get bored after a certain amount of homogeneous input. Unlike computers, brains are moody (well, unlike most computers, anyway), and the same input doesn't always produce the same output. And yet, brains are also smart and great at pattern recognition. The example I mentioned above about doing 1.0+2.0 a million times can be completed in exactly the same time as just once thanks to memoization. You could even try to fool it by doing 1.0+--2.0 and the result will also be computed only once. And olny a bairn cloud mkae snesn of tihs. It wloud taek a hlel of a lot of pminrrgamog to get a cpomtuer to udtersnand this.

Humans are the only self-aware animals alive right now (that we know exists; not to rule out aliens or anything because I strongly believe they exist) AFAIK.

http://en.wikipedia.org/wiki/Mirror_test#Animals_that_pass_the_mirror_test

[chrisname]

@Duoas
Wow. Calm down.

I see that instead of addressing my points one by one, and disproving them where you disagreed; you took my whole argument as a whole and turned it into an insult. Essentially, you used the same method as a child. Nice one. I have an excuse - I'm less than half your age, my mind hasn't developed. Of course I'm going to be ignorant and stupid. But you, being an adult, are supposed to show better judgment.

If by this:

I can't debate with amateur philosophers.

you are referring to this:

Here's something for you. Essentially, everything that lives, lives to make more of itself. I mentioned, to a friend, a type of fly that lives for only a few hours -- just long enough to reproduce. It doesn't even eat or drink. It's born, it lays eggs, then dies. His response was "What is the point of that?"

What is the point of anything? All you do in life leads up to one thing - IRL fork() [(that was a joke)].

that's just something I thought would be found interesting by anyone who hadn't already though of it. It was not an opportunity for you to decide I think I'm some kind of Aristotle in the making or something and then attack me for it.

I also see you ignored my attempt at friendly conversation. Please don't talk to me again. I don't like talking to adults that act like children.

@helios
Too much input, my brain got bored.

Actually, in that regard brains are exactly like computers. The difference is in how they process the data, and I'm not just talking about differences in chemical and electrical processes. Computers use chiefly linear processes, with very few parts parallelized, while brains are almost fully parallel, if that statement makes any sense.

If by that, you mean that computers process everything linearly (that is, as it appears (like when streaming data?)) that's not strictly true, is it? CPUs can process instructions out-of-order to (I assume) process them more efficiently or something. Unless I misunderstand you.

The example I mentioned above about doing 1.0+2.0 a million times can be completed in exactly the same time as just once thanks to memoization.

Actually, yeah. I didn't think of that -- once a human had calculated 1.0+2.0, if you asked them again they'd just recall it from memory. Although I think computers will definitely be able to do that. Perhaps they could store in ROM the result of every calculation they've ever performed? Now that'd probably take too much memory; but hell... they make 1 TiB hard disks now, when twenty years ago they were making 1.4 MiB floppy disks.

And olny a bairn cloud mkae snesn of tihs. It wloud taek a hlel of a lot of pminrrgamog to get a cpomtuer to udtersnand this.

Lol. I read somewhere that as long as the first and last letters are in the right places, it can still be made sense of (at least in English, I don't know if it works in other languages). Then again, I think I read that in a signature on a forum somewhere (I'm so glad this forum doesn't have signatures)...

Edit: A computer could understand some of that. If it was parsed word by word and checked (in the same way search engines remove spelling errors, I think it's called CBR or something; anyway I think they check each word against a database of correctly spelled words and the closest match is suggested as the correct answer), maybe it would eventually get the correct sentence using some kind of brute force algorithm. It'd probably spew about twenty different matches though. That's something a human couldn't do. I couldn't sit there and process that string into every possible match. Well, I could; but I'd be bound to make mistakes and I'd get bored. Also, it would take me HOURS. It'd probably take a computer a few minutes at most to do that...

Although, I guess a computer wouldn't really "understand" that. Computers are dumb. They just do things... that's probably the one thing they won't ever overcome -- they'll never truly understand what you're telling them to do, they'll just do it.

[mcleano]

@chrisname & Duoas ... Keep the peace!

Perhaps they could store in ROM the result of every calculation they've ever performed?

I don't think that could ever work efficiently. Firstly, would it not take to much time and processing power to find the answer. First they would have to recall the sum. Then recall the answer that goes with it. Not to mention the fact that 1.0+2.0 can also be written as 2.0+1.0. So it might spend all that time looking for a sum that it has an answer to but as it's written differently, you never find it and so result in computing it again.

[chrisname]

@chrisname & Duoas ... Keep the peace!

Not to cause mayhem, but I took that as an insult. Of course I'm going to try to defend myself.

I disagree,

Firstly, would it not take to much time and processing power to find the answer

I don't know. I don't think so. If this ROM was stored very close to the CPU (not on the die like registers, but like RAM) then possibly not. I can't find any information for the average time it takes to access memory for a 386 (and I can't imagine there is such info, with memory clock frequencies ranging from something like 200MHz (DRR) to 2000 MHz (DDR3 a la Patriot). There's also timings to take into account, which tend to go from something like 4-4-4-12 (lowest I've seen for DDR2) to 9-9-9-24 (DDR3) and FSB speed and a whole host of other stuff) but eventually it could work.

Not to mention the fact that 1.0+2.0 can also be written as 2.0+1.0. So it might spend all that time looking for a sum that it has an answer to but as it's written differently, you never find it and so result in computing it again.

Oh yeah.

[helios]

CPUs can process instructions out-of-order

I said chiefly linear. There are small few things that are parallelized, which is what allows for example pipelining in a CPU. However, the functioning as a whole is basically a single linear thread.

(Method to clean garbled textual input through brute force.)

Sure, a human wouldn't sit through that, but a human wouldn't have to, either. We're already equipped with the algorithms necessary to do it efficiently. We don't need to test every permutation of "udtersnand" to know what it means (for starters, we don't necessarily know that's how the text was garbled).

EDIT: Memoization at the CPU level is still more expensive than computing the answer again, no matter how you set it up. It only gets efficient when you memoize complex operations.

[chrisname]

I said chiefly linear. There are small few things that are parallelized, which is what allows for example pipelining in a CPU. However, the functioning as a whole is basically a single linear thread.

It's annoying actually that it does that. Is the optimization really worth it? To calculate clock frequency, I had to do a cpuid before each rdtsc because otherwise I would get even less accurate results (the CPU can't pipeline cpuid apparently. Or XOR, because that wouldn't make sense).

Memoization at the CPU level is still more expensive than computing the answer again, no matter how you set it up. It only gets efficient when you memoize complex operation

Well then, maybe the CPU could time the amount of time it took to do an operation. If it's above a threshold (say, 300 ns?), then the CPU stores it as a lengthy operation that shouldn't be tried again? Or should I quit while I'm ahead..?

Also the way it would be stored would not be as an exact sum.
For example, if tan(sin(cos(cos^-1(sin^-1(tan^-1(1.0 + 2.0)))))) was too long (it takes less than 3 seconds on my calculator, so I can't imagine it would take more than a few nanoseconds on a 386 or more powerful CPU) then we would store that somewhere, but not as that exact thing (although in this example (it wasn't a good idea to use something like this) it would have to be) and the CPU would search for any similarities.

Actually, I think I will quit while I'm ahead :)

[helios]

Is the optimization really worth it?

Um... Yes?

[chrisname]

It is?

Actually; I just searched instruction pipelining.

Wikipedia wrote:

A non-pipeline architecture is inefficient because some CPU components (modules) are idle while another module is active during the instruction cycle. Pipelining does not completely cancel out idle time in a CPU but making those modules work in parallel improves program execution significantly.

[...]

Advantages of Pipelining:

1. The cycle time of the processor is reduced, thus increasing instruction issue-rate in most cases.
2. Some combinational circuits such as adders or multipliers can be made faster by adding more circuitry. If pipelining is used instead, it can save circuitry vs. a more complex combinational circuit.

Disadvantages of Pipelining:

1. A non-pipelined processor executes only a single instruction at a time. This prevents branch delays (in effect, every branch is delayed) and problems with serial instructions being executed concurrently. Consequently the design is simpler and cheaper to manufacture.
2. The instruction latency in a non-pipelined processor is slightly lower than in a pipelined equivalent. This is due to the fact that extra flip flops must be added to the data path of a pipelined processor.
3. A non-pipelined processor will have a stable instruction bandwidth. The performance of a pipelined processor is much harder to predict and may vary more widely between different programs.

[hamsterman]

@chrisname
Please note that you are only writing your statements without any real arguments and without any doubt of their correctness.

And no, animals aren't self-aware. Humans are the only self-aware animals alive right now

,

animals don't have complex thoughts

Here's a fine example of what I just said. You don't know these things. You can only assume them from some primitive tests, yet this is only assumption. Science cannot give any more than that. All our knowledge is based on tests and assumptions. In psychology these assumptions are even more fragile becouse of the weak understanding of how does thinking really work. What I mean to say is that you should spend more time doubting complexity of your own human thoughts than denying that of other beings. (for example start with defining what is love you mentioned before and identify, what is there that animals don't have, or how many thoughts does an ameba have despite its ability to react to environment).
Also, my problem with these sentences is the word 'animal'. What differs us from animals? I believe that this is only a matter of complexity. I agree that humans are more advanced, but you cannot draw a solid line between humans and animals.

Computers are dumb. They just do things...

Again here. Do you do something more?.. As a programer you see how basic instructions assemble into complex applications. What prevents you from believing that similar system can produce something as complex as consciousness.

Please take my criticism with no offence and forgive my poor English...

Edit: lol you guys change topics realy quickly. Or maybe I just write so slowly...