As far as I''m aware (don''t quote me on this) the human brain is completely incapable of generating a sequence of truly random numbers either (just look at the proportion of people who, when asked to pick a number between one and ten, pick seven, or when asked to pick a vegetable choose a carrot).

An AI can only simulate Love because of one very simple reason. It is all in our mind.

The irrationality comes from an upset hormone balance in your brain. Why is it unbalanced? You (or rather, your brain) came to the completely rational choice that you like a person. It triggers several compeltely normal, non-random reaction. Including a chemical reaction that makes you "like" a person. The irrational acting is just a side effect. All you are trying to do is impress the person. The irrational part is just there, because you head to your "goal" with no regard for other things.

Who cares if you look stupid, as long as you think it works to win the girl for you? :D It is all about desires...

And, my PDA-AI with as an ultimate goal to assist me, well, that would be just he same as me with the ultimate goal to create true AI. Does that put me in an infinite loop as well?

Something similar was discussed on flipcode a while back, check out the thread here:
http://www.flipcode.com/cgi-bin/msg.cgi?showThread=RandomnessInGames&forum=discuss&id=86

That was about whether computers could create art, but the arguments are essentially the same. I''ll just repeat what I said there in this thread, cause it applies here as well:

Actually, if your John Searle, you''re pretty sure machines can''t be concious (he wrote "What computers can''t do" and "What computers still can''t do"). I don''t think his argument holds though, but if you''re interested in this kind of thing it''s good to look up his main arguments. The problem with saying that computers can''t do certain things, like feeling love, or creating art or writing sonnets is that it relies on somewhat strange assumptions. Obviously we, as humans are capable of these things, so there has to be some fundamental difference between the way we do it, and the way computers (could/will) do it. The argument for the fundamental difference often goes something like this:

All computers do is execute logical operations.
Executing logical operations can''t create art/feelings/conciousness
If the operations don''t do this, the system as a whole can''t do it.

The problem is that then either each part of humans has to be able to create art (creative molecules/neurons whatever) or there''s some kind of soul outside the world as we know it that is the source for all the art, love and conciousness (the final fantasy approach). If you replaced all the neurons of a human brain with electronic implants that functioned exactly as the original neurons, would the person with the new electronic brain cease to feel love and no longer create art? If he still does those things, then isn''t this an example of a computer/machine creating art?

*sigh*... still no one understands. Of course you can program a computer to express love or making it believe it loves someone. But to the AI that''s only variable with a value (ok lots of ''em variables =) but still it could be a bit saying
LovePerson02232 = 1
You can''t, atleast not today, program the computer to actually love the person. Look at Star Trek it took Data about 20-30 years (don''t remember) until he mastered his emotional chip. Ok ok it''s entirly fiction but on the other hand are''nt Qubit Quantum Bioligical computers with human mind and spirit that also??

// Shadows
Rogue Software

As far as AI that is closest the the brain goes, neural networks are an amazing concept. Basically model the brain using code and structures. They have been used in pattern recongnition, and can interpret human speech better than humans can in some cases. In addition, they are used heavily in weather and stock market prediction. I''m doing a project on them now. Although they take a while to train, like the human brain, they are capable of association and finding solutions to problems that don''t have a set, algorithmic approach.

As a reply to the fact that humans are''nt random and as evidence of that states that a lot of people pick 7 between 1 and 10 or carrot if asked to think about a vegetable.

Now someone didn''t think to far before posting that, I stated that computers cannot be random. Anyone can figure out that if you randomly choose a number between 1 to 10 lets say a million times. One of two of these numbers will be more used than the others because if all numbers were equally used that would be a chance in what 10 billion, and most important that would be more of a pattern like computer ai systimer type of way to generate random numbers. Humans can be random take this for an example: Close your eyes, don''t think about your fingers and press some numeric keys:
98645343574864312468743198435468734326343212132141
Now if this isn''t a random number then... what is?

// Shadows
Rogue - The EH-CGRPG Soon to awaken... - Software

In my defence - no, I didn't think too much before making that point, but I felt that it was a point that needed to be made. Furthermore, closing your eyes not thinking about your fingers and typing random keys on the numeric keyboard will NOT give you a truely random sequence of numbers. As far as I am aware there will always be some form of repeating patterns.

Take a look at your example. Very rarely in your sequence is the difference between two consecutive numbers greater than 2. This indicates some degree of dependance between the individual digits.

Doing a quick test shows that we would expect a truely random sequence of digits to have an average difference between consecutive numbers of 3.30 to 3 s.f. Your example has an average of 1.92 to 3 s.f. Although I haven't bothered to test for mathematical significance it certainly looks to me as though your sample sequence was not truely random.

This was the point that I was trying to make. I am unaware of any way in which humans are capable of generating truely random numbers. I may be wrong - please feel free to correct me if I am

EDIT: Slight numerical inaccuracy

Edited by - Enigma on November 8, 2001 3:19:46 PM

i read somewhere that a computer can generate random numbers based on something outside the computer (i found a website, whose URL i unfortunately forgot, that generated numbers by taking digital photographs of a lava lamp and performing a checksum on the colors, with some additional mathematical stuff). another way has something to do with reading the number of particles released from a radioactive isotope over a very small time scale.
granted, this isn''t useful as far as home computer users go, but i imagine that if you had the funding and wanted to use truly random numbers for something in your AI, you could set up something like that and get them.
i apologize for not remembering the links to any of this information.

--- krez (krezisback@aol.com)

Enigma: Ok, first of all I just want to apoligize for my rude behaviour, right then I was kinda pissed off and just needed something to shed my irritation at... no hard feelings?

I still disagree on the subject though ... Just because a number isn't 'jumpy' that doesn't make it any more random. The definition, or rather my definition, of random is simply something picked without thinking. Just because 987654321 is more organized some might say, it really isn't any more random than 196752348. As long as I don't think about it. But sure how about "tossing a coin" or "throwing a dice" isn't that atleast truly random? Now some might disagree and say that it is the coin or the dice that is random and not the human. I disagree though a coin can't be random heads or tail is simply a matter of how hard and at which angle you throw (wind, resistance, some jackass stopping the coin in mid air, blablabla).

As to the point of generating random numbers from a lava lamp or a radioactive isotope and derive that through various mathematical equations. Now I'm most definatly not a whiz on this subject but isn't that very much like taking a number from the systimer and derive that through various mathematical equations? Is a lavalamp random or is there a pattern? The same goes for the isotope... 'cause the timer sure as hell isn't =)

// Shadows
Rogue - The EH-CRPG soon to awaken... - Software

Edited by - shadows on November 8, 2001 4:36:18 PM

quote:

---

Original post by Shadows
As to the point of generating random numbers from a lava lamp or a radioactive isotope and derive that through various mathematical equations. Now I''m most definatly not a whiz on this subject but isn''t that very much like taking a number from the systimer and derive that through various mathematical equations? Is a lavalamp random or is there a pattern? The same goes for the isotope... ''cause the timer sure as hell isn''t =)

---

it is similar, except that you can get the value of the system timer, and thereby predict the sequence of numbers. with a lavalamp (or the radioactive isotope), you cannot predict what the seed will be (yes, a lava lamp is random and unpredictable; you know it will have various globs moving around, but it is impossible to tell the exact formation and color variations of those globs, which is what the algorithm uses). thus it will not be predictable at all ever. isn''t that the definition of random?
someone was saying that generating a sequence of random numbers should give you a bunch of numbers that are within a certain variance, et cetera... actually, random numbers don''t have to fit ANY criteria... they are random! you could roll a dice 100 times and get 100 ones, it is still random.
hmm... i was going to post a link to the LavaRand website (the one with the lava-lamp generated random numbers), which explained this much better than i ever could... but it is gone now... there is a new one but they aren''t done yet... a google search has some stuff about it by REAL mathematicians, since you apparently don''t believe it when i say it...

by the way, i practiced a lot and can control whether the coin lands on heads or tails


--- krez (krezisback@aol.com)