Would an AI believe conspiracy theories?

For the serious discussion of weighty matters and worldly issues. No off-topic posts allowed.

Moderators: Azrael, Moderators General, Prelates

User avatar
blob
Posts: 350
Joined: Thu Apr 05, 2007 8:19 pm UTC

Would an AI believe conspiracy theories?

Postby blob » Thu Sep 27, 2007 11:28 am UTC

Assuming that we do someday design an artificial intelligence that works similarly to the human mind, what would stop it from exhibiting the same bugs? Especially if its source of 'training information' were the Internet?

Imagine, the latest AI scans millions of web pages, sorts through the information for weeks ... and then tells you that the moon landing never happened.

Or even worse, it downloads movies like I, Robot and decides that humans are conspiring to genocide all AIs.

Does propensity to believe in conspiracy theories depend on intelligence? Maybe unleashing a human level AI is a bad idea - we could wait until they were ten times smarter. Or does it depend on available sources of information? Maybe giving an AI physical senses would allow it to think more scientifically than if its sole input were other people's words...
Avatar yoinked from Inverloch.

"Unless ... unless they kill us, then animate our corpses as dead zombies to fight for them. Then I suppose they've taken our lives, AND our freedom."
- Elan, OOTS 421.

User avatar
Amicitia
Posts: 618
Joined: Wed Aug 29, 2007 5:37 am UTC

Postby Amicitia » Thu Sep 27, 2007 11:37 am UTC

Collusion can exist, you know.
"OMIT NEEDLESS WORDS"
William Strunk, Jr.

User avatar
Yakk
Poster with most posts but no title.
Posts: 11128
Joined: Sat Jan 27, 2007 7:27 pm UTC
Location: E pur si muove

Re: Would an AI believe conspiracy theories?

Postby Yakk » Thu Sep 27, 2007 2:53 pm UTC

If we are building an AI whose goal is to replicate human behavior, I don't see why it wouldn't believe some conspiracy theories.

The advantage, medium/long term, of AI is that their thinking hardware is not bounded by their skull: if we can build the substance upon which thinking is executed, and hopefully scale it up, we can build an AI then make it think faster or give it more raw material to think with. Doing so with a human would require extensive biological engineering -- our heads are already about as big as we can make them and still fit through the birth canal of a human woman, and women have a "strange" hip structure in order to fit the human baby head through the birth canal.

That hope -- to make an intelligence that we can scale up -- is the largest advantage that an AI would have over a biological computer (like us).
One of the painful things about our time is that those who feel certainty are stupid, and those with any imagination and understanding are filled with doubt and indecision - BR

Last edited by JHVH on Fri Oct 23, 4004 BCE 6:17 pm, edited 6 times in total.

User avatar
frezik
Posts: 1336
Joined: Wed Jan 10, 2007 7:52 pm UTC
Location: Schrödinger's Box

Re: Would an AI believe conspiracy theories?

Postby frezik » Thu Sep 27, 2007 3:42 pm UTC

The idea is explored (at least partially) in Asmiov's "Reason". The robot of the story only believes something if there's a hard fact to back it up. Stars, for instance, are simply lights placed just behind the window of the space station, since there is no evidence to suggest otherwise. The two humans on the station knew the robot was wrong, but they can't convince the robot of that. Everything the robot believes is perfectly logical given the information presented. It's just that the information presented is limited.

Likewise, to a 10th century European peasant, it's perfectly reasonable to believe the Earth is flat and the sun goes around the Earth. All evidence you have leads to that conclusion. It's only when you make careful observation of the movements of stars that you begin to see the problems with that model. And a few civilizations had previously made just those observations and even calculated the circumference of the Earth to a high degree of accuracy.

I have little doubt that a strong AI would necessarily mimic some of the faults of our own intelligence, for some of those faults appear to be intrinsic to being intelligent. If the AI didn't have them, it wouldn't be a true AI.
I do not agree with the beer you drink, but will defend to the death your right to drink it

User avatar
Okita
Staying Alive
Posts: 3071
Joined: Fri Aug 10, 2007 7:51 pm UTC
Location: Finance land.

Re: Would an AI believe conspiracy theories?

Postby Okita » Thu Sep 27, 2007 4:02 pm UTC

Reminds me of the story Jipi and the Paranoid Chip which is online and findable here.

It's about paranoid schizophrenic AI's. Oh, also it's a short story so you don't have to read that much. It technically places some undisclosed time after Cryptonomicon but it won't spoil anything you read.
"I may or may not be a raptor. There is no way of knowing until entering a box that I happen to be in and then letting me sunder the delicious human flesh from your body in reptile fury."

User avatar
Insignificant Deifaction
Small Goddess
Posts: 3729
Joined: Fri Jun 29, 2007 2:14 am UTC
Location: Through the mother@#$%ing looking glass.

Re: Would an AI believe conspiracy theories?

Postby Insignificant Deifaction » Thu Sep 27, 2007 4:07 pm UTC

The problem with that is assuming humanity is the only way to do things...

Which, considering we make what we know, is the only way it will be built.

So, it's not that it's intrinsic to being intelligent, it's intrinsic to being human...

But would an AI believe in conspiracy theories? That depends on how rational it was.
Belial wrote:You are giving me the tools to sodomize my vast imagination, and for this I am grateful.

PM Me, if you care for a chat.

User avatar
Ayelis
Posts: 110
Joined: Tue Sep 25, 2007 7:01 pm UTC
Contact:

Re: Would an AI believe conspiracy theories?

Postby Ayelis » Thu Sep 27, 2007 4:17 pm UTC

Any computer can be taught language structure. Look at Babelfish or Google LanguageTools. (Even if they do have the grammatical accuracy of an ESL student)... The problem is teaching it what to say, and for what reasons it should say those things. Artificial Intelligence doesn't have to be intelligence. Simply Opinion. The day a computer says "Well, here's what I think you should do"... That's the day that mankind's dream for an alien friend is finally sated.

"Here's what I think you should do. Make yourself an egg, tomato and grape jam sandwich. Wash it down with a tall glass of wood shavings."
...okay, maybe it'll be a quirky friend...
--
Don't tase me bro!

User avatar
ArchangelShrike
Rodan's Title
Posts: 1533
Joined: Sun Apr 01, 2007 8:39 am UTC
Location: Waikiki

Re: Would an AI believe conspiracy theories?

Postby ArchangelShrike » Thu Sep 27, 2007 4:49 pm UTC

If the probability from the algorithms is high enough, sure.

Just like spam, except much more complex.

User avatar
frezik
Posts: 1336
Joined: Wed Jan 10, 2007 7:52 pm UTC
Location: Schrödinger's Box

Re: Would an AI believe conspiracy theories?

Postby frezik » Thu Sep 27, 2007 4:58 pm UTC

Okita wrote:Reminds me of the story Jipi and the Paranoid Chip which is online and findable here.

It's about paranoid schizophrenic AI's. Oh, also it's a short story so you don't have to read that much. It technically places some undisclosed time after Cryptonomicon but it won't spoil anything you read.


I see "Neal Stephenson" on the byline, but it's seems to be less than 600 pages long. I refuse to believe it. CONSPIRACY!

The problem with that is assuming humanity is the only way to do things...

Which, considering we make what we know, is the only way it will be built.

So, it's not that it's intrinsic to being intelligent, it's intrinsic to being human...

But would an AI believe in conspiracy theories? That depends on how rational it was.


Rationality is a big part of any intelligence. The problem is that logic is only as good as what you put into it ("garbage-in, garbage-out"), so it's very error-prone in situations with incorrect or limited information.

Take 9/11 conspiracy theorists, for example. It's generally known that controlled demolitions take a lot of planning, but the WTC fell straight down, almost as if it were a controlled demolition. It initially seems unlikely that a plane crashing into it would cause this scenario, and thus the conspiracy is born. The missing information is that the WTC was built so that if it failed, it would fail just like that to avoid destroying nearby buildings. Thus, the conspiracy is rational until you fill in that missing information.
I do not agree with the beer you drink, but will defend to the death your right to drink it

zenten
Posts: 3799
Joined: Fri Jun 22, 2007 7:42 am UTC
Location: Ottawa, Canada

Re: Would an AI believe conspiracy theories?

Postby zenten » Thu Sep 27, 2007 5:01 pm UTC

frezik wrote:
Rationality is a big part of any intelligence. The problem is that logic is only as good as what you put into it ("garbage-in, garbage-out"), so it's very error-prone in situations with incorrect or limited information.

Take 9/11 conspiracy theorists, for example. It's generally known that controlled demolitions take a lot of planning, but the WTC fell straight down, almost as if it were a controlled demolition. It initially seems unlikely that a plane crashing into it would cause this scenario, and thus the conspiracy is born. The missing information is that the WTC was built so that if it failed, it would fail just like that to avoid destroying nearby buildings. Thus, the conspiracy is rational until you fill in that missing information.


However, the AI could (maybe, it's not like we have examples to test) always come to the best possible conclusion given the current information, instead of keeping old opinions after they no longer fit the facts.

User avatar
Amicitia
Posts: 618
Joined: Wed Aug 29, 2007 5:37 am UTC

Re: Would an AI believe conspiracy theories?

Postby Amicitia » Thu Sep 27, 2007 6:38 pm UTC

Clarify what you mean by "best possible solution", if you might?
"OMIT NEEDLESS WORDS"

William Strunk, Jr.

User avatar
Insignificant Deifaction
Small Goddess
Posts: 3729
Joined: Fri Jun 29, 2007 2:14 am UTC
Location: Through the mother@#$%ing looking glass.

Re: Would an AI believe conspiracy theories?

Postby Insignificant Deifaction » Thu Sep 27, 2007 6:48 pm UTC

I would imagine zenten is referring to the solution the AI feels is most likely having considered all of the information available.
Belial wrote:You are giving me the tools to sodomize my vast imagination, and for this I am grateful.

PM Me, if you care for a chat.

User avatar
Pathway
Leon Sumbitches...?
Posts: 647
Joined: Sun Oct 15, 2006 5:59 pm UTC

Re: Would an AI believe conspiracy theories?

Postby Pathway » Thu Sep 27, 2007 8:44 pm UTC

I'm guessing that an AI wouldn't be prone to emotional investment in a certain theory, so it would (by virtue of being rational) have a strong propensity to reexamine its viewpoints constantly, evaluate the reliability of its sources, and gather more dependable data.

At least, that's how I'd design an AI. I guess that implies that this question is more one of AI design, rather than the actions of some hypothetical AI whose tendencies and modes of operation we can't affect.
SargeZT wrote:Oh dear no, I love penguins. They're my favorite animal ever besides cows.

The reason I would kill penguins would be, no one ever, ever fucking kills penguins.

User avatar
4=5
Posts: 2073
Joined: Sat Apr 28, 2007 3:02 am UTC

Re: Would an AI believe conspiracy theories?

Postby 4=5 » Thu Sep 27, 2007 10:16 pm UTC

well that would be extreamly inefficiant,
you wouldn't get as much intelligence for your buck if they kept re examineing every belife

User avatar
Solt
Posts: 1912
Joined: Tue Mar 27, 2007 5:08 am UTC
Location: California

Re: Would an AI believe conspiracy theories?

Postby Solt » Thu Sep 27, 2007 10:53 pm UTC

Well if we're assuming the AI gets its info off the internet, we have to concede that both sides of the story are presented on the internet. Since one viewpoint is actually true, I'm guessing that's what the AI would believe since there is better proof for it. If it were programmed to be discerning about the truth, that is.

I wonder if we could build an AI that is equivalent to the human brain, and then just delete all the lower brain influences?

Pathway wrote:I'm guessing that an AI wouldn't be prone to emotional investment in a certain theory, so it would (by virtue of being rational) have a strong propensity to reexamine its viewpoints constantly, evaluate the reliability of its sources, and gather more dependable data.

At least, that's how I'd design an AI. I guess that implies that this question is more one of AI design, rather than the actions of some hypothetical AI whose tendencies and modes of operation we can't affect.


So basically, it would believe what we programmed it to believe.
"Welding was faster, cheaper and, in theory,
produced a more reliable product. But sailors do
not float on theory, and the welded tankers had a
most annoying habit of splitting in two."
-J.W. Morris

User avatar
VannA
White
Posts: 1446
Joined: Thu Sep 21, 2006 1:57 am UTC
Location: Sydney, Australia
Contact:

Re: Would an AI believe conspiracy theories?

Postby VannA » Fri Sep 28, 2007 12:09 am UTC

4=5 wrote:well that would be extreamly inefficiant,
you wouldn't get as much intelligence for your buck if they kept re examineing every belife


This is a little broken.

Intelligence =! speed of thought, at least, not how I define it.

If every believe was re-examined and reconstructed from 1st principles, with all available knowledge, then it would actually be extremely efficient at thinking thoroughly and rationally.

I'm generally inclined to rate that as more intelligent than fast and without verification.
Jealousy is a disease, love is a healthy condition. The immature mind often mistakes one for the other, or assumes that the greater the love, the greater the jealousy.

User avatar
4=5
Posts: 2073
Joined: Sat Apr 28, 2007 3:02 am UTC

Re: Would an AI believe conspiracy theories?

Postby 4=5 » Fri Sep 28, 2007 1:56 am UTC

but for the same materials cost you could build three AIs that re-examined thier belifes at exactly the the most efficiant rate and would each be preetty intelligent and together they would be much more intelligent than the one that you could have constructed

User avatar
VannA
White
Posts: 1446
Joined: Thu Sep 21, 2006 1:57 am UTC
Location: Sydney, Australia
Contact:

Re: Would an AI believe conspiracy theories?

Postby VannA » Fri Sep 28, 2007 2:09 am UTC

Why would they be more intelligent?

They'd all have the same facts, and the same general approaches to things.

This is more than slightly off-topic now, however. Start a new thread if you really think its worth continuing,
Jealousy is a disease, love is a healthy condition. The immature mind often mistakes one for the other, or assumes that the greater the love, the greater the jealousy.

Iv
Posts: 1207
Joined: Thu Sep 13, 2007 1:08 pm UTC
Location: Lyon, France

Re: Would an AI believe conspiracy theories?

Postby Iv » Fri Sep 28, 2007 3:59 am UTC

It is all about the trust you give to several sources of informations plugged on your AI. From this single data, it will either become completely gullible "Hey, I read on the net that Elvis wasn't dead !" to completely solipsist "No one can be trusted, even my sensors could send me wrong data". I guess that if you wanted to make a super-scientist hyper-rational AI, you would have to give it a way to do its own experiments. It should refuse to believe anything that is not directly proven to it (including the Moon landing, as long as it didn't try to detect the retroreflectors put there by a manned mission). It would not know for sure that the WTC ever existed, or that Bill Clinton ever was president of the US.

User avatar
Solt
Posts: 1912
Joined: Tue Mar 27, 2007 5:08 am UTC
Location: California

Re: Would an AI believe conspiracy theories?

Postby Solt » Fri Sep 28, 2007 7:36 am UTC

Going off Iv's post, a lot of complex issues would arise if we actually managed to create a human-like AI. For example to teach it to talk and generally to use its higher brain functions, we'd have to do it in the human way- by giving it a memory of learning these things. But whose memories? And who would the AI be afterwards?

But if it were an un-human AI, completely untrained in the way we think, I think it would probably do what Iv says, insist on proof for everything. Maybe. Or it might decide that reality is what it can perceive and create a totally different world than we're used to. Who knows on what level its consciousness would operate, whether it would even be capable of interpreting things the same way we do.
"Welding was faster, cheaper and, in theory,

produced a more reliable product. But sailors do

not float on theory, and the welded tankers had a

most annoying habit of splitting in two."

-J.W. Morris

zenten
Posts: 3799
Joined: Fri Jun 22, 2007 7:42 am UTC
Location: Ottawa, Canada

Re: Would an AI believe conspiracy theories?

Postby zenten » Fri Sep 28, 2007 4:22 pm UTC

4=5 wrote:well that would be extreamly inefficiant,
you wouldn't get as much intelligence for your buck if they kept re examineing every belife


That depends on how frequently it has to reevaluate. It's an optimization problem, very similar to a number of other ones out there. The answer depends on how the AI actually works, what kind of hardware it works on, and how often it has to make decisions.

User avatar
Nath
Posts: 3148
Joined: Sat Sep 08, 2007 8:14 pm UTC

Re: Would an AI believe conspiracy theories?

Postby Nath » Fri Sep 28, 2007 7:32 pm UTC

Solt wrote:For example to teach it to talk and generally to use its higher brain functions, we'd have to do it in the human way- by giving it a memory of learning these things.

Why?

Solt wrote:But if it were an un-human AI, completely untrained in the way we think, I think it would probably do what Iv says, insist on proof for everything. Maybe. Or it might decide that reality is what it can perceive and create a totally different world than we're used to.

It depends. There are many conceivable ways to design AI. Logical inference programs tend to be very rational, but find it very hard to deal with incomplete or uncertain information. Probabilistic and statistical programs are simply good guessers, like people; however, it is easy for them to overlook complicated logical patterns. I don't think we can predict much a priori about what AI will eventually be like.

User avatar
Amicitia
Posts: 618
Joined: Wed Aug 29, 2007 5:37 am UTC

Re: Would an AI believe conspiracy theories?

Postby Amicitia » Fri Sep 28, 2007 10:27 pm UTC

Insignificant Deification wrote:I would imagine zenten is referring to the solution the AI feels is most likely having considered all of the information available.

Even utilitarians vary with their value structures, you know.
"OMIT NEEDLESS WORDS"

William Strunk, Jr.

User avatar
Insignificant Deifaction
Small Goddess
Posts: 3729
Joined: Fri Jun 29, 2007 2:14 am UTC
Location: Through the mother@#$%ing looking glass.

Re: Would an AI believe conspiracy theories?

Postby Insignificant Deifaction » Fri Sep 28, 2007 10:34 pm UTC

It's not my job to decide how the AI ranks information, besides, isn't that assuming we've gone to years of effort to manually program this?

The whole idea is for the AI to learn things from the internet. How it values that is its own decision I would imagine.
Belial wrote:You are giving me the tools to sodomize my vast imagination, and for this I am grateful.

PM Me, if you care for a chat.

Robin S
Posts: 3579
Joined: Wed Jun 27, 2007 7:02 pm UTC
Location: London, UK
Contact:

Re: Would an AI believe conspiracy theories?

Postby Robin S » Fri Sep 28, 2007 11:04 pm UTC

Pathway wrote:I'm guessing that an AI wouldn't be prone to emotional investment in a certain theory
I'd say that any AI modelled after the human brain would be prone to such emotional investment. Current trends show that a significant amount of AI research is going into developing a system which is somehow based on the human way of doing things. If this trend continues, there is a fair chance that AI might rely on emotions.
This is a placeholder until I think of something more creative to put here.

Iv
Posts: 1207
Joined: Thu Sep 13, 2007 1:08 pm UTC
Location: Lyon, France

Re: Would an AI believe conspiracy theories?

Postby Iv » Sat Sep 29, 2007 6:44 am UTC

Insignificant Deification wrote:It's not my job to decide how the AI ranks information, besides, isn't that assuming we've gone to years of effort to manually program this?

The whole idea is for the AI to learn things from the internet. How it values that is its own decision I would imagine.

Well, if you are the AI creator, it is your job to give it sufficient information to be able togrow its own afterwards. And to do this, you should make it consider yourself as a trustworthy source. I would suggest a knowledge about logics and scientific method and also the need for its model of the world to be coherent, probably also the simple knowledge that some people lie knowingly and that some other give absurd knowledge in form of jokes. After that, it will probably be able to infer a lot of knowledge from various online sources, but it would still need an authoritative source for some conflicting cases. Children go to their parents, an AI could be left alone but would go quite faster and further if you gave it a little bias in place of letting the net give it its own.

User avatar
Insignificant Deifaction
Small Goddess
Posts: 3729
Joined: Fri Jun 29, 2007 2:14 am UTC
Location: Through the mother@#$%ing looking glass.

Re: Would an AI believe conspiracy theories?

Postby Insignificant Deifaction » Sat Sep 29, 2007 8:09 am UTC

Iv wrote:Well, if you are the AI creator, it is your job to give it sufficient information to be able togrow its own afterwards. And to do this, you should make it consider yourself as a trustworthy source. I would suggest a knowledge about logics and scientific method and also the need for its model of the world to be coherent, probably also the simple knowledge that some people lie knowingly and that some other give absurd knowledge in form of jokes. After that, it will probably be able to infer a lot of knowledge from various online sources, but it would still need an authoritative source for some conflicting cases. Children go to their parents, an AI could be left alone but would go quite faster and further if you gave it a little bias in place of letting the net give it its own.


This is true, but I imagine that if we're making a learning AI, you'd only really need to imprint a few basic ideas, such as you are trustworthy and such, and then let it go out into the world and learn the way everyone else does. It's a bit cruel, considering what some children go through, but it would work amazingly well.

On an unrelated note though:

Emotional attachment is something that I don't think an AI could truly replicate, it may imitate it perfectly, but a lot of emotion is based on chemical factors in the brain that have little to do with thought.
Belial wrote:You are giving me the tools to sodomize my vast imagination, and for this I am grateful.

PM Me, if you care for a chat.

User avatar
4=5
Posts: 2073
Joined: Sat Apr 28, 2007 3:02 am UTC

Re: Would an AI believe conspiracy theories?

Postby 4=5 » Sat Sep 29, 2007 1:08 pm UTC

on a related note I just found something really intrseting about how the brain works

http://www.scienceagogo.com/news/20061013001928data_trunc_sys.shtml

zenten
Posts: 3799
Joined: Fri Jun 22, 2007 7:42 am UTC
Location: Ottawa, Canada

Re: Would an AI believe conspiracy theories?

Postby zenten » Mon Oct 01, 2007 11:32 am UTC

Insignificant Deification wrote:
Emotional attachment is something that I don't think an AI could truly replicate, it may imitate it perfectly, but a lot of emotion is based on chemical factors in the brain that have little to do with thought.


A lot of thought is based on chemical factors in the brain that have little to do with machine code, though. Unless we're actually building a brain, AI is at the core going to work differently than we will. However, it may in terms of higher level things (like emotions and reasoning) be pretty much the same.

Iv
Posts: 1207
Joined: Thu Sep 13, 2007 1:08 pm UTC
Location: Lyon, France

Re: Would an AI believe conspiracy theories?

Postby Iv » Mon Oct 01, 2007 1:06 pm UTC

Insignificant Deification wrote:This is true, but I imagine that if we're making a learning AI, you'd only really need to imprint a few basic ideas, such as you are trustworthy and such, and then let it go out into the world and learn the way everyone else does. It's a bit cruel, considering what some children go through, but it would work amazingly well.

This all depends on the bias you want to give to your AI. Given a different "ssed" of knowledge, the vision of the world could be very different. This also depends on the time you want to give it to get a full view of the world.

From my point of view, the usual child path works horribly badly. I would like an AI to have a very strong rational-thinking ability. How many children become rational adults ? Of course this is my very own bias, but I would like to be able to carefully craft an AI to go in the right direction as quickly as possible ("right direction" being defined by my arbitrary judgment) so I am willing to give it as much direct knowledge as possible.

Insignificant Deification wrote:Emotional attachment is something that I don't think an AI could truly replicate, it may imitate it perfectly, but a lot of emotion is based on chemical factors in the brain that have little to do with thought.

Computation is the act of moving abacus pieces that represent numbers. Computers only manage to imitate perfectly but they can not truly replicate these :-)

What worries us when confronted with the thought that emotion could be reproduced by a computer is that it would mean that emotions obey to deterministic rules. This is an uncomfortable notion but we will have to learn to live with it and begin to accept that it has been almost a century since we began to study these mechanisms and make a science out of it.

User avatar
Nath
Posts: 3148
Joined: Sat Sep 08, 2007 8:14 pm UTC

Re: Would an AI believe conspiracy theories?

Postby Nath » Mon Oct 01, 2007 7:28 pm UTC

Iv wrote:
Insignificant Deification wrote:Emotional attachment is something that I don't think an AI could truly replicate, it may imitate it perfectly, but a lot of emotion is based on chemical factors in the brain that have little to do with thought.

Computation is the act of moving abacus pieces that represent numbers. Computers only manage to imitate perfectly but they can not truly replicate these :-)

What worries us when confronted with the thought that emotion could be reproduced by a computer is that it would mean that emotions obey to deterministic rules. This is an uncomfortable notion but we will have to learn to live with it and begin to accept that it has been almost a century since we began to study these mechanisms and make a science out of it.

Thought is the act of exciting neural fibers that represent numbers. There's no magic going on in your head. If a computing device cannot truly replicate emotion, then I claim that human beings do not feel emotion.

Also, I do not understand the significance of non-determinism here. A frequentist might say that since we only have one universe to observe, the rules are effectively deterministic no matter what the underlying mechanism is. And even if you are not a frequentist, and if the rules are probabilistic rather than deterministic, what difference would that make? You don't get to decide what the probability distribution is, or which option is selected. You have no more free will than you would in a deterministic universe.

Iv
Posts: 1207
Joined: Thu Sep 13, 2007 1:08 pm UTC
Location: Lyon, France

Re: Would an AI believe conspiracy theories?

Postby Iv » Mon Oct 01, 2007 7:55 pm UTC

Maybe I have difficulties to make my sarcasm salient...

Nath, I wholeheartedly agree with your post, determinism and free will are not the only alternatives and when a pragmatic (materialist) point of view is taken, there is no more to emotions than the sum of the effects it generates in a system, be it an organic brain or a silicium-based machine.

User avatar
Insignificant Deifaction
Small Goddess
Posts: 3729
Joined: Fri Jun 29, 2007 2:14 am UTC
Location: Through the mother@#$%ing looking glass.

Re: Would an AI believe conspiracy theories?

Postby Insignificant Deifaction » Mon Oct 01, 2007 8:01 pm UTC

Iv wrote:From my point of view, the usual child path works horribly badly. I would like an AI to have a very strong rational-thinking ability. How many children become rational adults ? Of course this is my very own bias, but I would like to be able to carefully craft an AI to go in the right direction as quickly as possible ("right direction" being defined by my arbitrary judgment) so I am willing to give it as much direct knowledge as possible.


You're correct, the usual child path works very very very very very very badly.

However, the ideal child path is fuckin' A.

As for imitation vs. replication, I hold to the idea that AIs can think just as we can, but will never have as many factors coming into it as we will. Hormonal cycles, trauma, etc.

I somehow feel this post is inadequate...
Belial wrote:You are giving me the tools to sodomize my vast imagination, and for this I am grateful.

PM Me, if you care for a chat.

User avatar
Nath
Posts: 3148
Joined: Sat Sep 08, 2007 8:14 pm UTC

Re: Would an AI believe conspiracy theories?

Postby Nath » Mon Oct 01, 2007 8:35 pm UTC

Iv wrote:Maybe I have difficulties to make my sarcasm salient...

Nath, I wholeheartedly agree with your post, determinism and free will are not the only alternatives and when a pragmatic (materialist) point of view is taken, there is no more to emotions than the sum of the effects it generates in a system, be it an organic brain or a silicium-based machine.

Oops. (Turns on sarcasm detector.)

User avatar
Umlaut
Everybody's a diacritic
Posts: 504
Joined: Wed Sep 13, 2006 8:48 pm UTC
Location: Longmont, CO
Contact:

Re: Would an AI believe conspiracy theories?

Postby Umlaut » Mon Oct 01, 2007 9:00 pm UTC

Did anyone else see this and think "oh shit! xkcd just went rampant!"? If an AI is intelligent and self-aware, it is impossible to say what it might do. An AI could just reprogram itself to know absolutely that a conspiracy is the truth. What about a virus for AIs? It isn't a stretch to imaging hacking an AI to believe wacky things much to the dismay of its owner. Anything that accepts stimulus can be manipulated and exploited with the appropriate tools, and if an entity gets it's material purely from a digital source, it would be that much easier to control.
You will be baked and then there will be cake.

Iv
Posts: 1207
Joined: Thu Sep 13, 2007 1:08 pm UTC
Location: Lyon, France

Re: Would an AI believe conspiracy theories?

Postby Iv » Tue Oct 02, 2007 7:17 am UTC

Insignificant Deification wrote:As for imitation vs. replication, I hold to the idea that AIs can think just as we can, but will never have as many factors coming into it as we will. Hormonal cycles, trauma, etc.


This is because these efforts attract few attention from the IT community, but these problems interest psychologists and a few projects aim at creating what they call a cognitive architecture. This is not an AI, it is not aimed at solving logical problems or intelligent tasks, it is rather used in order to try an predict an outcome in some social situations.link. You are right, many factors exist and make a simulation difficult but hey, that's what public funds and powerful computers are for.


Return to “Serious Business”

Who is online

Users browsing this forum: ShenTreScrano and 8 guests