Tables of random numbers have probably been used for multiple purposes at least since the Industrial Revolution. The first published table appears to be by the English statistician L.H.C. Tippett. In the digital age, algorithmic random number generators have largely replaced these tables. The NIST Randomness Beacon expands the use of randomness to multiple scenarios in which the latter methods cannot be used. The extra functionalities stem mainly from three features. First, the Beacon-generated numbers cannot be predicted before they are published. Second, the public, time-bound, and authenticated nature of the Beacon allows a user application to prove to anybody that it used truly random numbers not known before a certain point in time. Third, this proof can be presented offline and at any point in the future. For example, the proof could be mailed to a trusted third party, encrypted and signed by an application, only to be opened if needed and authorized.
I've long held this belief, although I have not phrased it with such enjoyable humor.
When I heard people say things similar to " 'they' say that in 20 years, computers will be smarter than people - do you think that's true?" I reply with "they're already 'smarter', and artificial intelligence already exists".
I follow up with the assertion that when AI is, if it's not already, a fully developed reality it won't take the form we're expecting, much in the same way that "artificial flight" did not take the form of humans flying like birds.
I still think that when claiming computers are 'smarter' than people, you are side-stepping the fundamental problem with which this article revolves around, and that is the problem of properly defining 'intelligence'. Can computers write a classic novel? Can computers compose a masterpiece? Can computers perform in a ballet? Sure, they can replicate automated tasks and calculate much faster than humans, but that does not make them inherently smarter, especially when you consider that humans created computers, not the other way around.
Yes, I absolutely agree with you! In they're already 'smarter', I had smarter in single quotes - which I imagined myself saying with finger-air quotes. Computers are "smarter" than us in a very narrowly defined way only. As you pointed out, properly defining intelligence is very difficult.
Starts out funny, but gets painful when the author starts trying to defend his claim.
The article as a whole never defines what "human intelligence" is -- which enables the author to use the "no true Scotsman fallacy" whenever a counterexample is brought up (see the comments).
> "science knows that we make decisions before the rational parts of our brains activate"
How true is this? Always? Sometimes? Who did this study? What were the controls?
> math skills are real, for example. But a computer can do math.
Doesn't cover all math skills -- just the easy ones. Using math != creating math. Could a computer create, say, set theory or proof theory?
> Language skills are real too, but a computer can understand words and sentence structure.
Again, the example doesn't cover all human language skills -- just the easy ones.
Maybe the biggest problem is the implicit assumption the author makes that everything that humans do depends on human intelligence. If you make that assumption, and it turns out to be false -- if humans do lots of things that don't require human intelligence -- then, obviously, it's very easy to cherry-pick examples of things humans do that don't require human intelligence. But that doesn't have anything to do with whether human intelligence exists or not, or whether we can duplicate it with computers.
5 comments
[ 2.1 ms ] story [ 20.5 ms ] threadI go to government issued random number site.
http://www.nist.gov/itl/csd/ct/nist_beacon.cfm
Tables of random numbers have probably been used for multiple purposes at least since the Industrial Revolution. The first published table appears to be by the English statistician L.H.C. Tippett. In the digital age, algorithmic random number generators have largely replaced these tables. The NIST Randomness Beacon expands the use of randomness to multiple scenarios in which the latter methods cannot be used. The extra functionalities stem mainly from three features. First, the Beacon-generated numbers cannot be predicted before they are published. Second, the public, time-bound, and authenticated nature of the Beacon allows a user application to prove to anybody that it used truly random numbers not known before a certain point in time. Third, this proof can be presented offline and at any point in the future. For example, the proof could be mailed to a trusted third party, encrypted and signed by an application, only to be opened if needed and authorized.
https://www.fourmilab.ch/hotbits/secure_generate.html
When I heard people say things similar to " 'they' say that in 20 years, computers will be smarter than people - do you think that's true?" I reply with "they're already 'smarter', and artificial intelligence already exists".
I follow up with the assertion that when AI is, if it's not already, a fully developed reality it won't take the form we're expecting, much in the same way that "artificial flight" did not take the form of humans flying like birds.
The article as a whole never defines what "human intelligence" is -- which enables the author to use the "no true Scotsman fallacy" whenever a counterexample is brought up (see the comments).
> "science knows that we make decisions before the rational parts of our brains activate"
How true is this? Always? Sometimes? Who did this study? What were the controls?
> math skills are real, for example. But a computer can do math.
Doesn't cover all math skills -- just the easy ones. Using math != creating math. Could a computer create, say, set theory or proof theory?
> Language skills are real too, but a computer can understand words and sentence structure.
Again, the example doesn't cover all human language skills -- just the easy ones.
Maybe the biggest problem is the implicit assumption the author makes that everything that humans do depends on human intelligence. If you make that assumption, and it turns out to be false -- if humans do lots of things that don't require human intelligence -- then, obviously, it's very easy to cherry-pick examples of things humans do that don't require human intelligence. But that doesn't have anything to do with whether human intelligence exists or not, or whether we can duplicate it with computers.