IS Forum
Forum Index Register Members List Events Mark Forums Read Help

Go Back   International Skeptics Forum » General Topics » Science, Mathematics, Medicine, and Technology
 


Welcome to the International Skeptics Forum, where we discuss skepticism, critical thinking, the paranormal and science in a friendly but lively way. You are currently viewing the forum as a guest, which means you are missing out on discussing matters that are of interest to you. Please consider registering so you can gain full use of the forum features and interact with other Members. Registration is simple, fast and free! Click here to register today.
Reply
Old 9th September 2021, 07:33 PM   #1
Cainkane1
Philosopher
 
Cainkane1's Avatar
 
Join Date: Jul 2005
Location: The great American southeast
Posts: 8,947
How close are to to having a self aware, self programming machine?

Comuputers are great but I wish I were young enough to see the day when we can speak with, and interact with a self aware machine. Perhaps a robot.
__________________
If at first you don't succeed try try again. Then if you fail to succeed to Hell with that. Try something else.
Cainkane1 is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 9th September 2021, 07:51 PM   #2
arthwollipot
Observer of Phenomena
Pronouns: he/him
 
arthwollipot's Avatar
 
Join Date: Feb 2005
Location: Ngunnawal Country
Posts: 73,386
Why? What purpose would such a device serve?
__________________
We are all #KenBehrens
arthwollipot is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 9th September 2021, 08:18 PM   #3
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Actually, I'd say we passed that point quite some time ago. Not in any way that's useful for anything else, but we did.

Self-awareness is easy. In fact, it's easier for a machine than for a human. If there's anything that can be fully aware of its state, traits and programming, it's for a computer. Not in any consciousness sense, mind you, but it can know itself -- including, yes, that it's aware of its surroundings and even how -- to a degree that you can't.

Self programming? That's been possible ever since we first stored program code in read-write memory. It's been more of a bad practice to avoid, but it's been possible.

The most trivial is if you use Prolog as a programming language. It's a language that's both:

A. based on rules, not an explicit program sequence. To write, say, a factorial function, you just give it the rules of how it works. Sorta (in natural language, not actual Prolog syntax)

factorial(0)=1
factorial(i)=i*factorial(i-1)

Then when you ask it what's the factorial of 3, it just figures it out from those rules. Better yet, it can use them in reverse too. Like, with the above rules you can ask it something like what's the number whose factorial is 6, and it will tell you it's 3.

B. has no distinction between code and data. A program can add or change those rules on the fly.

So, anyway, between those two, it's been used exactly for machine learning back in the day. The program would learn by basically adding such knowledge (i.e., rules) to its own code.

So there we go, not only it's been done, but was quite trivial. I wrote such a program back in high school.

And again, it being mapped to the same space as data, it meant that it would be fully "aware" of what rules it's working by.

We gave up on that because binary rules proved to go nowhere in AI learning. We use bayesian learning these days, which turned out to actually work for a change. But it being more useless for actual AI than vertical ejection seats in a helicopter, is a different issue from whether it exists
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 9th September 2021 at 08:20 PM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 9th September 2021, 11:34 PM   #4
Roger Ramjets
Philosopher
 
Roger Ramjets's Avatar
 
Join Date: Jun 2008
Posts: 5,549
Originally Posted by arthwollipot View Post
Why? What purpose would such a device serve?
It would satisfy our desire to have a slave dear friend who tireless serves and laughs at our jokes without secretly wanting to strangle us. One that we cherish and admire, but when the need arises can be thrown under the bus without remorse because it's just a machine.
__________________
We don't want good, sound arguments. We want arguments that sound good.
Roger Ramjets is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 12:27 AM   #5
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by Roger Ramjets View Post
It would satisfy our desire to have a slave dear friend who tireless serves and laughs at our jokes without secretly wanting to strangle us. One that we cherish and admire, but when the need arises can be thrown under the bus without remorse because it's just a machine.
Nailed it.

This is why people get dogs.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 12:35 AM   #6
dann
Penultimate Amazing
 
Join Date: Feb 2004
Posts: 14,948
Originally Posted by Roger Ramjets View Post
It would satisfy our desire to have a slave dear friend who tireless serves and laughs at our jokes without secretly wanting to strangle us. One that we cherish and admire, but when the need arises can be thrown under the bus without remorse because it's just a machine.
Your description sounds like Trump's relationship with other people, except that the ones he throws under the bus aren't machines.
I don't think those of us who are mentally stable adults (unlike very stable geniuses) have that desire. I'm fine with my friends telling me I'm being a jerk if they think I am. That's what friendship is about. Friendship is not a one-sided narcissistic cult.
__________________
/dann
"Stupidity renders itself invisible by assuming very large proportions. Completely unreasonable claims are irrefutable. Ni-en-leh pointed out that a philosopher might get into trouble by claiming that two times two makes five, but he does not risk much by claiming that two times two makes shoe polish." B. Brecht
"The abolition of religion as the illusory happiness of the people is required for their real happiness. The demand to give up the illusion about its condition is the demand to give up a condition which needs illusions." K. Marx
dann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 12:47 AM   #7
bluesjnr
Professional Nemesis for Hire
 
bluesjnr's Avatar
 
Join Date: Jun 2009
Location: Home.
Posts: 8,893
Originally Posted by Mike Helland View Post
Nailed it.

This is why people get dogs.
I need to re-programe mine then. They've reversed the roles.
__________________
bluesjnr is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 12:50 AM   #8
Shrinker
Graduate Poster
 
Shrinker's Avatar
 
Join Date: Oct 2004
Posts: 1,070
Careful what you wish for. When the days comes you can have an intelligent peer-to-peer conversation with a computer, 18 months later, V2 will be out and you'll be beneath it's contempt.
Shrinker is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 01:21 AM   #9
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by bluesjnr View Post
I need to re-programe mine then. They've reversed the roles.
If your dog laughs at your jokes and would throw you under a bus, you have a cat, and you ingested LSD.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 05:08 AM   #10
erlando
Master Poster
 
erlando's Avatar
 
Join Date: Apr 2007
Posts: 2,001
If you by "self-aware" mean "conscious" then the answer is "very, very far". We might not even get there with current technology.
__________________
"If it can grow, it can evolve" - Eugenie Scott, Ph.D Creationism disproved?
Evolution IS a blind watchmaker
erlando is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 06:05 AM   #11
Bikewer
Penultimate Amazing
 
Bikewer's Avatar
 
Join Date: Sep 2003
Location: St. Louis, Mo.
Posts: 13,007
A couple of months ago, I would have been in the “not for quite a while” camp. But Neil DeGrasse Tyson’s “Star Talk” had an episode:

https://youtu.be/lljmkWEsH4w

With a fellow (Josh Clark) who said that the breakthroughs are “imminent” and that we could be seeing such machine intelligence in the very near future.

I know the debate over this is intense, and it’s a popular concept (both for good and evil) in science fiction. But I think that we’ll have to decide how to treat such a thing when (inevitably) it appears.
Bikewer is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 10th September 2021, 09:36 AM   #12
Wudang
BOFH
 
Wudang's Avatar
 
Join Date: Jun 2003
Location: People's Republic of South Yorkshire
Posts: 13,804
I hate to post video links but this is this guy's speciality. I saw him at Barnsley Skeptics In The Pub and he knows his stuff. 1st video


https://www.robots.ox.ac.uk/~nickh/public_engagement/
__________________
"Your deepest pools, like your deepest politicians and philosophers, often turn out more shallow than expected." Walter Scott.
Wudang is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 12th September 2021, 08:53 PM   #13
arthwollipot
Observer of Phenomena
Pronouns: he/him
 
arthwollipot's Avatar
 
Join Date: Feb 2005
Location: Ngunnawal Country
Posts: 73,386
Originally Posted by erlando View Post
If you by "self-aware" mean "conscious" then the answer is "very, very far". We might not even get there with current technology.
I think that if it can be done, it will, for no other reason than to prove that it can be done. However, there is no utility in such a device. There is no existing problem that it solves. It has no function. If it asks you "Why am I here?" the only possible answer is "We made you just to see if we could."
__________________
We are all #KenBehrens
arthwollipot is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 12th September 2021, 09:17 PM   #14
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Originally Posted by erlando View Post
If you by "self-aware" mean "conscious" then the answer is "very, very far". We might not even get there with current technology.
Luckily enough, though, "self-aware" doesn't mean "conscious." Not the least, lucky because we don't really have definition of "consciousness" that actually works for anything else than philosophy debates. I'd also point out that even those usually fail to capture the meaning used by the layman, which is basically a vague and nebulous "whatever the heck makes me more special than my cat." Most of the definitions of consciousness fail to do that, by just requiring some form of being able to experience sensations OR be aware that stuff exists, depending on whose definition you use. In the process, they allow even a Roomba to qualify as conscious, since it's aware of its surroundings and has sensors that allow it to 'feel' when it bumped into a table leg. That's what tends to happen when you try to move from an undefined "whatever I can claim to be, in order to feel special" to trying to actually have a rigorous definition of exactly what is required.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 12th September 2021 at 09:20 PM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 12th September 2021, 09:28 PM   #15
Skeptic Ginger
Nasty Woman
 
Skeptic Ginger's Avatar
 
Join Date: Feb 2005
Posts: 90,508
Since we don't have a good understanding of how we ourselves perceive consciousness, it's a long long way off.

AIs can be programed with all sorts of algorithms, including some that allow complex decision making which might appear like some kind of independent intelligence. But it simply is not self-awareness or actual consciousness. It's mimicry and nothing more.

When we figure out how mammals experience consciousness, then we'll see if it can be artificially replicated.
Skeptic Ginger is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 12th September 2021, 10:28 PM   #16
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by Skeptic Ginger View Post
Since we don't have a good understanding of how we ourselves perceive consciousness, it's a long long way off.

AIs can be programed with all sorts of algorithms, including some that allow complex decision making which might appear like some kind of independent intelligence. But it simply is not self-awareness or actual consciousness. It's mimicry and nothing more.

When we figure out how mammals experience consciousness, then we'll see if it can be artificially replicated.
How can we be sure we haven't already?

One thing we can say about the brain is that it's a model of reality, inside reality.

If there's panpsychism going on, and all things have "being", than a model of reality has being.

In that case anything that is a dynamic model of reality, has some kind of consciousness.

It's not so much that you and I are conscious beings, but we contain models of reality within reality.

The universe is conscious of itself through those models of itself.

A self driving car does the same thing.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 03:32 AM   #17
Puppycow
Penultimate Amazing
 
Puppycow's Avatar
 
Join Date: Jan 2003
Location: Yokohama, Japan
Posts: 26,195
Originally Posted by Shrinker View Post
Careful what you wish for. When the days comes you can have an intelligent peer-to-peer conversation with a computer, 18 months later, V2 will be out and you'll be beneath it's contempt.
Anyone see the movie Ex Machina?

Kinda frightening!

(I'll put the rest in spoiler tags.)

It's one of several recent fictional films or TV shows dealing with the moral issues around a self-aware intelligent robot. The robot is smart enough to question its own captivity, and doesn't like the power that its creator has over her.

I'm sure that these robots (at least some of them) would be designed to be (willing) sex slaves. But what if they were smart enough to question this situation? Would it be ethical to not give them full and equal human rights and the freedom of self-determination?
__________________
A fool thinks himself to be wise, but a wise man knows himself to be a fool.
William Shakespeare
Puppycow is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 04:38 AM   #18
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
We're turning it into the other thread we already had some time ago, but basically let's just say this: the robot won't feel sad or happy or anything else, unless we program it to.

Even in humans it's actually not a result of analyzing your situation consciously, but of other chemical signals which get used in response to conditions that were useful to deal with from an evolutionary perspective. It's separate processing tasks, if you want to model them in a program. E.g., you get bored out of your skull if someone locks you in a closet until they next need you, just because evolution needed a signal to keep that animal doing SOMETHING for its survival, even if it's training. Not because you analyzed your situation at a conscious level and decided that boredom is the best course of action.

In a robot, you can program that as you wish. If it's, say, a maid robot (something like this: https://www.nexusmods.com/fallout4/mods/34739,) you can program it to be extremely happy when it can do chores for you. If it's an industrial bending robot, you can program it to be extremely happy when it can bend a metal bar to a given shape. If it's a glorified Roomba, you can make it experience happiness when it can get some dirt inside itself. Or whatever.

Making it self-aware would mean basically it being aware that it's aware, or that it experiences those feelings. (And, as I was saying, that step actually more trivial for a robot than for a human.) So basically my robo-maid would be aware that bringing me a nice cup of tea is making her happy. Would probably even be able to figure out that it's just because she's programmed that way.

But... so what? I mean I'm also fully aware that I like sex only because my nervous system is wired that way, but that doesn't make me want to swear it off.

But basically even cartoons have figured this one out, so I'm not sure why it's such a conundrum. I mean, when the superstitious robot in Futurama goes "I'll believe what I was programmed to believe", that's exactly what I'm talking about. It's self-aware, but so what? If he's not programmed to experience a cognitive dissonance unpleasantness about that belief, he just doesn't.

Or if we're talking movies, take the minions from Overlord. They're fully aware that they're created by a player, and that they're programmed to want to serve a player. Hell, Albedo is even aware that the player reprogrammed her to be madly in love with him. But even she's like, "So? What difference does it make?" And really, they got it right: if the AI isn't programmed to feel any discomfort about that, they won't.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 04:41 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 04:45 AM   #19
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
We're turning it into the other thread we already had some time ago, but basically let's just say this: the robot won't feel sad or happy or anything else, unless we program it to.
How would you do that?

And how would you know?

You seem to have reached your conclusion about what things can do, but only if we program it to.

What is that based on?

Do you have that much experience programming AIs?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 04:49 AM   #20
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
I may not be the best authority, but yes, I have enough experience with programming, some of it even with AI, to know that a program doesn't do anything else than what its source code says. Which may not be what you intended if you've got bugs, but it's still just exactly what the source code says. If you have a bug, you just wrote code that does something else than you intended, but what the system does is still just follow what the source code says.

If that source code doesn't produce an "I'm unhappy" state of some sort, then it just does not happen.

You don't even need to be an expert in AI or even programming to know that's not the case. You just need to have even the most basic understanding of programming. If you've ever written even a Hello World and understand why it will always produce "Hello World", and not go "Screw you, I'm bored of this Hello World nonsense" even in a trillion years, that's all the knowledge you need.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 04:53 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 04:54 AM   #21
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
If that source code doesn't produce an "I'm unhappy" state of some sort, then it just does not happen.

You don't even need to be an expert in AI or even programming to know that's not the case. You just need to have even the most basic understanding of programming and not be delusional.
I'm not convinced you've done a lot of AI.

What's an example of an AI you've developed and what does it do?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 04:56 AM   #22
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
And I'm saying that all you need to know is that if the source code doesn't produce a given state -- whether it's "I'm unhappy" or anything else -- then it doesn't happen.

A Turing machine is a finite state machine. It can't reach arbitrary extra states. To do so would mean that basically you can go through a graph and and reach a node that's not on that graph.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 04:58 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:00 AM   #23
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
And I'm saying that all you need to know is that if the source code doesn't produce a given state -- whether it's "I'm unhappy" or anything else -- then it doesn't happen.

A Turing machine is a finite state machine. It can't reach arbitrary extra states. To do so would mean that basically you can go through a graph and and reach a node that's not on that graph.
I'm not in doubt about how strongly you are convinced you are right.

That much is totally obvious.

Check this out:

https://techcrunch.com/2018/12/31/th...ppointed-task/

https://www.youtube.com/watch?v=Lu56xVlZ40M
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:06 AM   #24
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
And as your first will tell you, past the whole sensationalist narrative: "This is really just a lesson in the oldest adage in computing: PEBKAC. “Problem exists between keyboard and chair.”" I.e., it's purely a problem of the humans, not some program awareness. The computer doesn't feel anything, nor want anything, nor is even aware that it "cheated". It just solves the problems in the ways it's been programmed to.

In the case of a learning AI, it basically tries to figure out connections between things, and it's told whether it was right or wrong. That's how bayesian learning works, really. If you tell it that some way is consistently right, it doesn't know that it's cheating or whatever. It just keeps doing that. In the case in the article, it turns out that the only way it stumbled upon, which kept getting confirmed as producing results, was not the way its designers intended, and was using other data than they thought it was. It's not a case of the computer being lazy, or wanting to cheat, or really having any intent, it's just it not figuring out how to solve the problem in any other way, and going with what works.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 05:08 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:08 AM   #25
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
It just solves the problems in the ways it's been programmed to.
Patently false.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:09 AM   #26
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Originally Posted by Mike Helland View Post
Patently false.
Really? This time it's my turn to ask how you know that
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:11 AM   #27
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
Really? This time it's my turn to ask how you know that
Well, the video on the AI that broke the game gives plenty examples, but here's a pretty good example.

If it was programmed to solve problems in a certain way, why does it take several million runs to find new solutions?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:19 AM   #28
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
And I'm saying that I actually read and understood the paper linked in that article (which is here: https://arxiv.org/pdf/1712.02950.pdf) while you just seem to want to believe the sensationalist journalism parts which even your link says are false. And then going "patently false" without even understanding why. (Other than it contradicting your wild imagination.)

Even in the introduction they tell you that the program was only training 2 functions to satisfy specific constraints. Then those functions are applied to an image. It also helps if you understand how training an AI works, which, as I've said in message #24, is just Bayesian training.

THAT is what it was programmed to do, and yes, that is exactly what it did.

Incidentally that needs a lot of data to run on. That's how bayesian learning works. It needs a lot of data (which, yes, can amount to millions or even billions of runs) to refine its probabilities to something useful.

The program doesn't even understand WTH it is doing there, what those images represent, or anything. It's just training those two functions.

All that happened in the end is that they ended with functions that didn't work quite like they thought it would. In fact, they ended up with steganography.

But the program doesn't understand any of that. It just trained two functions, as programmed to do.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 05:22 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:29 AM   #29
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
The program doesn't even understand WTH it is doing there, what those images represent, or anything. It's just training those two functions.
Sounds about as human as anything I've heard of.

Did you watch this video?

https://www.youtube.com/watch?v=Lu56xVlZ40M

If what you are saying is true, it wouldn't invent new solutions after millions of run.

The source code would have just solved the problem in the way it was told to do.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:37 AM   #30
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
And I'm saying that what it was designed to do, the problem it was solving, was refining those two functions, by running them over millions of data sets. All modern AI is statistics. Which I already told you when I told you it's bayesian learning. Asking why a statistics program doesn't just spit the solution without needing all that data is what's "patently stupid."
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 05:39 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:37 AM   #31
Gord_in_Toronto
Penultimate Amazing
 
Gord_in_Toronto's Avatar
 
Join Date: Jul 2006
Posts: 21,175
If a machine says, "I'm conscious", how we would we know any different?

I'll toss this into the discussion:

A developer built an AI chatbot using GPT-3 that helped a man speak again to his late fiancée. OpenAI shut it down.

https://www.theregister.com/2021/09/..._openai_gpt_3/

It may be just Eliza on steroids, but so convincing.
__________________
"Reality is what's left when you cease to believe." Philip K. Dick
Gord_in_Toronto is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:41 AM   #32
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
And I'm saying that what it was designed to do, the problem it was solving, was refining those two functions, by running them over millions of data sets. Asking why a statistics problem doesn't just spit the solution without needing all that data is what's "patently stupid."

Code:
function add(x, y) {
     return x + y
}
That needs data (x and y) and produces a reproducible result.

Do you think that's what AI code looks like?

Do you think training AI's is deterministic?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:44 AM   #33
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Originally Posted by Mike Helland View Post
Code:
function add(x, y) {
     return x + y
}
That needs data (x and y) and produces a reproducible result.

Do you think that's what AI code looks like?

Do you think training AI's is deterministic?
Nobody said any of that, so I'm not sure how you even got to that idea. But I guess there's no limit to what one can misunderstand, when one doesn't know jack about what they're talking about
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:50 AM   #34
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
Nobody said any of that, so I'm not sure how you even got to that idea. But I guess there's no limit to what one can misunderstand, when one doesn't know jack about what they're talking about

Take a set of data and run your statistics algorithm on it.

Repeat that a million times.

Does it produce novel solutions?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:50 AM   #35
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Originally Posted by Gord_in_Toronto View Post
If a machine says, "I'm conscious", how we would we know any different?

I'll toss this into the discussion:

A developer built an AI chatbot using GPT-3 that helped a man speak again to his late fiancée. OpenAI shut it down.

https://www.theregister.com/2021/09/..._openai_gpt_3/

It may be just Eliza on steroids, but so convincing.
And that's kinda the whole point: it's just Eliza on steroids.

And it becomes even less impressive when you realize that some people have developed emotional connections even with the dumb old Eliza. And a lot larger a subset were convinced that it has some kind of intelligence that it just provably didn't have.

So, yes, try it on enough people, and you'll find one who's convinced he's talking to his late girlfriend
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 05:59 AM   #36
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Originally Posted by Mike Helland View Post
Take a set of data and run your statistics algorithm on it.

Repeat that a million times.

Does it produce novel solutions?
That's in fact how it works. In fact, since you insist on that youtube video, that's literally how it works, and they even tell and show you that it starts with small random movements and gradually zeroes in on something that works. That's how AI learning works. It tries various random stuff, and builds a probability matrix of what works, and then refines it in the next run, and so on.

THAT's why you need millions of runs. Because you're not just programming a "return x+y", you're building a probability matrix that eventually zeroes in on x+y. And you keep running it until that matrix has zeroed in on it.

Of course, that would involve actually understanding what the video is about.

The fact is, we programmers tend to use anthropomorphising language. We talk about the code trying stuff, wanting to do stuff, figuring out stuff, etc, expecting that the interlocutor is also clued enough to understand what we actually mean. Like that when I say that this function tries to figure out a way out of a maze (or in your video trying to get to a position where it has line of sight on the opposing team), the other guy also knows enough about programming to understand that actually the code doesn't literally do any of that. Like, that if it's a learning AI, I expect that the other guy would know it's actually it just dumbly runs a loop of crunching numbers, without any actual intent or real understanding.


In fact, your video is actually the best illustration of what I was talking about: "look at those happy eyes" doesn't mean the AI actors are actually happy or anything. It's just some animation that's triggered when a solution is found. It's just a programmed visual cue for the user, not actually being happy. It could just as well be programmed to show those little people crying when they found a solution, and they still would just keep trying to find a solution, as they're programmed to.

Conversely, unless you program them to look sad when they've lost, they won't.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 06:03 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 06:05 AM   #37
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
That's in fact how it works. In fact, since you insist on that youtube video, that's literally how it works, and they even tell and show you that it starts with small random movements and gradually zeroes in on something that works. That's how AI learning works. It tries various random stuff, and builds a probability matrix of what works, and then refines it in the next run, and so on.
I get that's how AI works.

You compared it to doing statistics.

That's not how you do statistics.
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 06:06 AM   #38
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
What in Lucifer Morningstar's good name do you even think bayesian anything is about, if not statistics?

Edit: I think I understand what's confusing you, though. It's not just some statistics of the problem. It's conditional probabilities that also depend on what you just did, and what you learned before (i.e., that probability matrix from the previous run.) As such, you actually need to actually do stuff and learn from it, it's not just running statistics on the input set and calling it a day.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?

Last edited by HansMustermann; 13th September 2021 at 06:09 AM.
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 06:09 AM   #39
Mike Helland
Master Poster
 
Join Date: Nov 2020
Posts: 2,206
Originally Posted by HansMustermann View Post
What in Lucifer Morningstar's good name do you even think bayesian anything is about, if not statistics?
Check this out:

https://www.calculator.net/statistic...30#inputfields

I hit the "calculate" button 3 times.

How many times do I have to hit it until it gives me a different answer?
Mike Helland is online now   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Old 13th September 2021, 06:11 AM   #40
HansMustermann
Penultimate Amazing
 
HansMustermann's Avatar
 
Join Date: Mar 2009
Posts: 19,189
Except, as I've said in that "Edit:" above, that's not the statistics that a learning AI calculates. It's not just statistics of the input image in your first link, or of the level in your second link, which yes, would not change. It's statistics of what you tried, and how much of that worked. You can't build those on just the input level. You have to actually run those millions of tries, to produce those statistics.
__________________
Which part of "Ph'nglui mglw'nafh Cthulhu R'lyeh wgah'nagl fhtagn" don't you understand?
HansMustermann is offline   Quote this post in a PM   Nominate this post for this month's language award Copy a direct link to this post Reply With Quote Back to Top
Reply

International Skeptics Forum » General Topics » Science, Mathematics, Medicine, and Technology

Bookmarks

Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Forum Jump


All times are GMT -7. The time now is 03:07 PM.
Powered by vBulletin. Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.

This forum began as part of the James Randi Education Foundation (JREF). However, the forum now exists as
an independent entity with no affiliation with or endorsement by the JREF, including the section in reference to "JREF" topics.

Disclaimer: Messages posted in the Forum are solely the opinion of their authors.