Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×

Comment Re:Every 30 days. (Score 1) 247

And some of the smaller websites will just mail your password to you on request. Just had one like that last week. Unbelievable in this day and age.

And another website I logged into last week, a frequent flyer program for a major airline, had a maximum of 6 numbers for the password. Not characters, numbers only! No idea how they store it, but something tells me they're probably not using MD5 and have never heard of salt.

Comment Re: writer doesn't get jeopardy, or much of anythi (Score 1) 455

Look at the complexity of today's computer chips. We stuff billions of logic gates into a square centimeter of silicon. Would it really be beyond our capabilities to make a copy of the structure of a human brain, but without all the blood and other biological nastiness, and make it orders of magnitude faster? We already pretty much understand how a neuron works, it's just the emergent behaviour of billions of those neurons connected to each other that still evades us. But all we need to do is build it and see what it does. I'm sure we will some day. I have no idea whether or not it will be truly "sentient" since we don't even know what that word means, but outperform us it certainly will.

Comment Re: writer doesn't get jeopardy, or much of anythi (Score 1) 455

Human brains appear so powerful because they take lots of shortcuts and make simplifications that are "close enough for government work". We are very good at discarding irrelevant data and making wild guesses. When asked to do a simple task but do it extremely accurately and repeatedly, we struggle. We are basically cheating all the time. Computers have vastly more power but are wasting most of it by being extremely precise. If we figure out how to let them compress their data in a usable, structured way, I think they probably do have the power to surpass us. The programming just isn't there yet. Also, they would need a lot more parallelism. All it would take, is someone using today's manufacturing techniques to build a chip with lots of interconnects, structured similar to a human brain. Computers switch millions of times faster than neurons (neurons get up to about 200 Hz max), so they'll outperform us pretty much immediately.

Really, look at how today's computers process an image. They look at every single pixel and make calculations on them trying to find basic structures. You try looking at a million numbers, given to you as one long list, and figuring out if it contains a picture of a car. The computer has that power, we just have to channel it in a different way.

Comment Re: writer doesn't get jeopardy, or much of anythi (Score 1) 455

"Already, computers are waaay more powerful than human minds"

no they aren't. Seriously, watch (...)

Let's use the same reasoning the other way around.

"Human minds are waaay more powerful than computers"

no they aren't. Seriously, watch a human solve a hard sudoku. These humans attempt to mimic basic computer tasks. They take something like an hour to do the calculations to fill in the symbols. And usually fail at that. It takes a computer roughly a tenth of a millisecond to do that same bit of calculation. And it never fails.

See? Like I said, we just haven't figured out yet how to steer all the power of computers towards actual intelligence. The human brain is good at parallelism (which computers currently still struggle with) but neurons fire at rates up to 200 Hz while computer circuits switch more than ten million times faster. They are already better at playing chess, long considered by many to be an impossible thing as it required "real intelligence" that would never be achieved by computers. They'll be driving cars soon (they already can in a very limited way). That, too, was considered impossible, how could a computer possibly process all that visual data? And whenever we manage to get them to perform some task (like flying an airplane, for example), they do so vastly more accurately than we do.

I'm sure that, once someone starts building chips that were specifically designed to have lots of interconnections structured similar to a human brain (instead of the current topology that still works more or less like a big switchboard), and we scale it up to the same number of nodes, it will immediately outperform our brains by orders of magnitude. And then imagine what kind of architectures that brain could come up with.

Comment Re: writer doesn't get jeopardy, or much of anythi (Score 1) 455

It could design the new hardware, which can then be manufactured. At some point the brains would be linked straight to the manufacturing equipment, so the chips could design and produce their successors. So at that point, yes, it could simply shit out better CPUs and plug them into itself. Of course it would have to be set up that way by humans initially, but from that point on...

Comment Re: writer doesn't get jeopardy, or much of anythi (Score 3, Interesting) 455

If you compare the power usage and performance of a Commodore 64 to today's laptops, I think we've done a pretty good job of exponentially increasing power efficiency. Already, computers are waaay more powerful than human minds, we just haven't figured out how to steer all this power towards actual intelligence. If mother nature can create human minds that function on a few sandwiches a day, I'm sure we'll be able to surpass that. Of course it can't continue to grow exponentially forever, but it can certainly scale well beyond the combined power of the seven billion human brains on this planet today.

Comment Re:Ignorance does not equal a new state (Score 1) 59

Just because you can't discern the difference between light and matter in this state -- this does not mean they are the same thing.

Exactly, if you just move to a different state that has different laws, and maybe even a different definition of pi, you'll find that it's perfectly legal to discern between light and matter.

Comment Re:Ask the credit card for a refund (Score 5, Interesting) 307

For my software, I've had chargebacks from people just saying they didn't want the software anymore, without any further explanation (after they had already received their registration code and could use the software with no restrictions). I was told there was nothing I could do. Instead of a $15 sale, I had to pay a $15 chargeback fee. All I could do was disable that registration code in future updates. Fortunately it was a relatively rare occurrence.

Slashdot Top Deals

One man's constant is another man's variable. -- A.J. Perlis

Working...