AI says God or No

[Edgarcito]

Thought this would be a interesting topic.....without too much contention.  Do we think AI would conclude there is a God or confirm a particular religion over others.

 

Interesting in my mind, but maybe not everyone's.

 

Thoughts if you would like to contribute....

 

Thx.

[walterpthefirst]

Do we think AI would conclude there is a God or confirm a particular religion over others?

 

 

No.

 

The belief in deities and gods is a function of human emotion. 

 

Because AI uses only on logic it can only arrive at logical explanations of reality.

 

A purely logical look at the evidence does not reveal the presence of any deities.

 

An emotional look at the evidence does reveal the presence of deities.

 

But that is because we want to see them - not because they are actually there.

 

 

 

Thank you,

 

Walter.

 

 

[Edgarcito]

1 hour ago, walterpthefirst said:

Do we think AI would conclude there is a God or confirm a particular religion over others?

 

 

No.

 

The belief in deities and gods is a function of human emotion. 

 

Because AI uses only on logic it can only arrive at logical explanations of reality.

 

A purely logical look at the evidence does not reveal the presence of any deities.

 

An emotional look at the evidence does reveal the presence of deities.

 

But that is because we want to see them - not because they are actually there.

 

 

 

Thank you,

 

Walter.

 

 

So you are saying that emotions are not essentially data sets that are incapable of being recognized?

 

Edit: Kind of like recognizing a particular pattern in the space time continuum but in a matrix of computer language.  Logic to define biology outside of biology itself.  Seems very plausible to me....

[walterpthefirst]

4 hours ago, Edgarcito said:

So you are saying that emotions are not essentially data sets that are incapable of being recognized?

 

Edit: Kind of like recognizing a particular pattern in the space time continuum but in a matrix of computer language.  Logic to define biology outside of biology itself.  Seems very plausible to me....

 

No, I'm following the exact remit of your question, Edgarcito.

 

Do we think AI would conclude there is a God or confirm a particular religion over others?

 

If you think that AI doesn't use logic to function then please present your evidence for this claim.

 

Otherwise your initial question sets down the remit (logic) of how AI must tackle the question.

 

Given that remit, all AI can do is look at the evidence in a purely logical way.

 

 

The question of AI recognizing emotions is a separate issue.

 

 

Thank you,

 

Walter.

 

 

 

 

[older]

I tried it and it would not take a stand. I can't remember exactly what it wrote, but the gist of it was that religion is not something that AI can determine, and that it is not ChatGPT's role to confirm or deny religion.

[mwc]

17 hours ago, Edgarcito said:

Thought this would be a interesting topic.....without too much contention.  Do we think AI would conclude there is a God or confirm a particular religion over others.

 

Interesting in my mind, but maybe not everyone's.

 

Thoughts if you would like to contribute....

 

Thx.

     You have to define AI.

 

     What we call AI is simply machine learning.  Simply put they're a set of algorithms that are trained or  "learn" from a given set of data.  There's no actual intelligence as we tend to think of it involved in the process.

 

     So if an AI were trained on a set of data that were biased towards gods being real then the AI would conclude that gods were real.  Furthermore if the data said that a particular god was real then that would be the one that the AI would conclude was real.  On the other hand if the data was biased in the direction then the AI would draw another conclusion.

 

     Now, of course, this is how humans tend to reason but the AI doesn't have the ability to seek out or include additional information like humans are able to do especially things like conducting experiments and the like to test its knowledge.

 

     Anyhow, a "General AI" is sort of the holy grail of AI's.  It would be something that could learn like a human.  It could gather information, weigh information, and reason given that information.  Again, perform all the tasks in a way like we do but in a way that is faster and utilizing much more information.  A general AI is something that is a very long ways off if not impossible to create (at least with our current technology).  It's not something to worry about.  What conclusions something like this might draw about a god is anyone's guess.  It might come to a conclusion there is a god but that god is human, since we created it, or is not human but some form of computer/AI and it created us in order to create it.  The possibilities are fairly endless since a general AI would amount to a hyper-intelligent individual and there's no way to know what it might think unless we control its knowledge (which means it's no longer a general AI).

 

     All the current generation of AI's are fun to play with and will get better but they're not going to really solve any of these problems.  They will be something that you will start to deal with in day-to-day living as I figure they'll replace low-wage support workers in the not to distant future.  Many jobs, such as front-line tech support and the like, will be lost to them.

 

          mwc

 

[◊ Krowb ◊]

12 hours ago, older said:

I tried it and it would not take a stand. I can't remember exactly what it wrote, but the gist of it was that religion is not something that AI can determine, and that it is not ChatGPT's role to confirm or deny religion.

 

That is a boundary line the developers wrote into the code to not weigh into religious matters.  If you structure the prompts correctly, you can get it to do all kinds of things. 

 

@mwc is spot on.  Current AI is machine learning and answers probabilistically based on what's in its dataset. If it's not in the dataset, you can't get the answer.  Further, your answer will vary over different sessions because it's pulling from different portions of the dataset over time.

 

GPT4 may be able to pass the UBE portion of a bar exam, but absolutely commits malpractice as there are state specific quirks that it misses or does the exact opposite of what particular state law requires. I tested with two states in which I'm licensed and GPT can't get it right as it's not trained on state specific law.  In both cases, the requested document would not conform to state law and would not be legally enforceable.  And this is a real danger because these AI state their answers so authoritatively.

[◊ Krowb ◊]

Also, want to clarify, if you ask it straight up to write a legal document you'll get a similar answer to what Older got, but with respect to "needing to find a licensed attorney in the jurisdiction".  But again, with some creative prompts, you can get it to generate what would be legal documents, only . . . it will be general law and be missing any state specific requirements or information.

[☆ DarkBishop ☆]

20 hours ago, Edgarcito said:

Thought this would be a interesting topic.....without too much contention.  Do we think AI would conclude there is a God or confirm a particular religion over others.

 

Interesting in my mind, but maybe not everyone's.

 

Thoughts if you would like to contribute....

 

Thx.

I would have to assume Walter is right. That AI unless programmed to "act" emotional would make a logical conclusion. What conclusion it would make would be based on available evidence. 

 

I'm not certain what the final conclusion would be though. We have a lot of data and we have made many great discoveries concerning our universal cosmology. But I don't believe we will ever be able to explain where the original material that created our universe came from. If it was a massive star that exploded sending atoms, neutrons, and protons out that formed all the elements. How did the star form? So "we" meaning humans and AI will never get all the answers to conclude definitively that there was never a creator. 

 

This is one reason I remain agnostic/non-theist rather than agnostic/atheist. I don't know if there is a God but if there is, it is not like anything theology has been able to come up with. 

 

As far as AI having emotions I think that would have to be programmed into it. Our emotions are affected by chemicals in our bodies. Hormones and endorphins that AI would not have. It would most likely draw a logical conclusion. Which is what makes the concept of AI troubling to us. There is logic to the movies like terminator. If we do develope AI and AI does become self aware who knows what it would decide about us?

 

DB

[Edgarcito]

4 hours ago, mwc said:

     You have to define AI.

 

     What we call AI is simply machine learning.  Simply put they're a set of algorithms that are trained or  "learn" from a given set of data.  There's no actual intelligence as we tend to think of it involved in the process.

 

     So if an AI were trained on a set of data that were biased towards gods being real then the AI would conclude that gods were real.  Furthermore if the data said that a particular god was real then that would be the one that the AI would conclude was real.  On the other hand if the data was biased in the direction then the AI would draw another conclusion.

 

     Now, of course, this is how humans tend to reason but the AI doesn't have the ability to seek out or include additional information like humans are able to do especially things like conducting experiments and the like to test its knowledge.

 

     Anyhow, a "General AI" is sort of the holy grail of AI's.  It would be something that could learn like a human.  It could gather information, weigh information, and reason given that information.  Again, perform all the tasks in a way like we do but in a way that is faster and utilizing much more information.  A general AI is something that is a very long ways off if not impossible to create (at least with our current technology).  It's not something to worry about.  What conclusions something like this might draw about a god is anyone's guess.  It might come to a conclusion there is a god but that god is human, since we created it, or is not human but some form of computer/AI and it created us in order to create it.  The possibilities are fairly endless since a general AI would amount to a hyper-intelligent individual and there's no way to know what it might think unless we control its knowledge (which means it's no longer a general AI).

 

     All the current generation of AI's are fun to play with and will get better but they're not going to really solve any of these problems.  They will be something that you will start to deal with in day-to-day living as I figure they'll replace low-wage support workers in the not to distant future.  Many jobs, such as front-line tech support and the like, will be lost to them.

 

          mwc

 

Thanks, that makes sense given my ignorance.  I think I was wanting the "general AI" conclusion without bias.  Uncertain what the task would entail to ask the AI without bias.....likely a difficult task off the cuff.

 

Thanks for the input.....and @Krowb as well.

[TheRedneckProfessor]

If human intelligence has not definitively concluded there is a god, why would an intelligence that humans developed conclude any differently?  If artificial intelligence is a product of human intelligence, and can only calculate based on the knowledge that human intelligence provides, then as concerns anything about which human intelligence is undecided, unconvinced, or uncertain, it seems that artificial intelligence could only mirror the same ambiguity.  Serious inquiry, here, @Edgarcito: why would it occur to you that something we create could do better than we have?

[Edgarcito]

13 hours ago, TheRedneckProfessor said:

  Serious inquiry, here, @Edgarcito: why would it occur to you that something we create could do better than we have?

I think AI likely has the ability to evaluate a data set and recognize trends or grouping we don't readily see...small to large.  Are our behaviors X or Y and then compare that to various doctrines.  Might give an insight into one religion being more certain than another...imo.

[walterpthefirst]

3 minutes ago, Edgarcito said:

I think AI likely has the ability to evaluate a data set and recognize trends or grouping we don't readily see...small to large.  Are our behaviors X or Y and then compare that to various doctrines.  Might give an insight into one religion being more certain than another...imo.

 

Which doctrines?

 

There are tens of thousands of variations of doctrine within Christianity and many thousands more within Islam, Judaism, Hinduism, Buddhism, etc.  

 

Unless you put the correct doctrine in you then the AI won't find the correct answer.

 

It will find a spurious one, just as mwc and Krowb have said.

 

If you don't ask the right question you won't get the right answer.

 

Garbage In = Garbage Out

 

 

 

 

[☆ DarkBishop ☆]

13 minutes ago, Edgarcito said:

I think AI likely has the ability to evaluate a data set and recognize trends or grouping we don't readily see...small to large.  Are our behaviors X or Y and then compare that to various doctrines.  Might give an insight into one religion being more certain than another...imo.

Edgarcito, how could one religion be more certain than another when none of them reflect reality or have any definitive proof that their God or belief is true? You don't think that AI would favor testable and provable information over emotional based biased religious doctrine?

 

DB

[☆ DarkBishop ☆]

9 minutes ago, walterpthefirst said:

 

Which doctrines?

 

There are tens of thousands of variations of doctrine within Christianity and many thousands more within Islam, Judaism, Hinduism, Buddhism, etc.  

 

Unless you put the correct doctrine in you then the AI won't find the correct answer.

 

It will find a spurious one, just as mwc and Krowb have said.

 

If you don't ask the right question you won't get the right answer.

 

Garbage In = Garbage Out

 

 

 

 

 

Are we talking about existing AI. Or the future possibilities of AI? I'm basing my thoughts on more advanced AI that will probably be around in the future. 

 

DB

[Edgarcito]

54 minutes ago, DarkBishop said:

Edgarcito, how could one religion be more certain than another when none of them reflect reality or have any definitive proof that their God or belief is true? You don't think that AI would favor testable and provable information over emotional based biased religious doctrine?

 

DB

You look at the information first and then assess independently to the continuity to a particular religion.  

[mwc]

1 hour ago, Edgarcito said:

I think AI likely has the ability to evaluate a data set and recognize trends or grouping we don't readily see...small to large.  Are our behaviors X or Y and then compare that to various doctrines.  Might give an insight into one religion being more certain than another...imo.

     Even if a perfect AI came to be.  If it came to the "wrong" conclusion about any religion(s) it would simply be ignored/denounced/refuted/etc. since this isn't what religion is all about.

 

     Machine learning (or colloquially AI), even in its current form, is helpful for doing some of the tasks you suggest but I'd consider it more a research assistant/tool than a researcher if we're going to look at it this way.  And AI will stay that way in ways of "faith" since no one invested in those matters wants something without a "soul" informing them on what is truth but it could be used to help them examine those things they already accept as truth.

 

     So your hypothetical unbiased AI could analyze historical documents and possibly help those involved in research find otherwise overlooked information but beyond that it's not of much use in this area.

 

          mwc

 

[walterpthefirst]

3 hours ago, DarkBishop said:

 

Are we talking about existing AI. Or the future possibilities of AI? I'm basing my thoughts on more advanced AI that will probably be around in the future. 

 

DB

 

I based my comments on Edgarcito's initial question, DB.

 

He didn't ask about future AI so I've confined my responses to him to current-day AI.

 

 

Thank you,

 

Walter.

[walterpthefirst]

3 hours ago, Edgarcito said:

You look at the information first and then assess independently to the continuity to a particular religion.  

 

Edgarcito,

 

 

As others have pointed out to you, AI can only do what it is programmed to do and nothing more.  So, if the parameters you want it to follow are flawed or ambiguous AI will not realize that you've made a mistake or have worded its instruction ambiguously.  It will just follow your flawed and ambiguous instructions.  Garbage in equals garbage out.  For example, your sentence, which I've quoted above would lead AI to select a religion on the basis of its continuity.  

 

https://en.wikipedia.org/wiki/Zoroastrianism

 

Zoroastrianism has been continuously practiced from pre-Christian (and possibly pre-Judaic) times right up until today.  That religion is therefore to logical one for AI to select, using your parameters.  And AI will only do what is logical, not what you think you want.

 

So, is this the answer that you wanted to hear?

 

 

 

Walter.

[☆ DarkBishop ☆]

Just gonna drop this in right here  

[TheRedneckProfessor]

11 hours ago, Edgarcito said:

I think AI likely has the ability to evaluate a data set and recognize trends or grouping we don't readily see...small to large. 

Doesn't human intelligence have the ability to evaluate data sets and recognize trends and groupings?  And yet, there is still no definitive conclusion of the existence of a god?

 

11 hours ago, Edgarcito said:

Are our behaviors X or Y and then compare that to various doctrines.

Human psychology is already providing us with answers to our behaviors.  Sometimes our behaviors are the result of certain doctrines (hence, indoctrination); other times, certain doctrines are the result of our behaviors.  Still, this brings us no closer to a conclusion. 

 

Yet, whatever information is fed into the AI concerning our behaviors would necessarily be filtered through human intelligence and behaviors.  Meaning the computer would only have access to information that we already know.

 

11 hours ago, Edgarcito said:

Might give an insight into one religion being more certain than another

More certain to whom?  Adherents to any particular religion already have the certainty required to accept it; while skeptics and detractors are unlikely to be convinced simply because a computerized algorithm produced the conclusion it was programmed to produce.  Which would bring us no closer to a conclusion about the existence of a god. 

[Edgarcito]

On 4/13/2023 at 9:33 PM, TheRedneckProfessor said:

Yet, whatever information is fed into the AI concerning our behaviors would necessarily be filtered through human intelligence and behaviors.  Meaning the computer would only have access to information that we already know.

 

More certain to whom?  Adherents to any particular religion already have the certainty required to accept it; while skeptics and detractors are unlikely to be convinced simply because a computerized algorithm produced the conclusion it was programmed to produce.  Which would bring us no closer to a conclusion about the existence of a god. 

I just believe there are many ways to look at information outside of "programmed to produce", an answer.  I gather we currently have means of deciding where things "happen" in the brain but probably not a picture or functionality of that happening.  So suppose you could describe a function in an array of 1s and 0s and then ask the AI to see where it may also identify that same array within it's database.  And if it does and is found in places where we wouldn't expect it to be found, then ask ourselves why. 

Is there a relationship between the similar arrays in two locations we wouldn't expect.  Kind of like an actual picture of a thought.  

 

Even in today's routine QC programs, people graph data because we may see a trend that we can't see in one form but maybe in another.  I think it likely would be the case in analyzing a computer process/function.

[walterpthefirst]

1 hour ago, Edgarcito said:

I just believe there are many ways to look at information outside of "programmed to produce", an answer.  I gather we currently have means of deciding where things "happen" in the brain but probably not a picture or functionality of that happening.  So suppose you could describe a function in an array of 1s and 0s and then ask the AI to see where it may also identify that same array within it's database.  And if it does and is found in places where we wouldn't expect it to be found, then ask ourselves why. 

Is there a relationship between the similar arrays in two locations we wouldn't expect.  Kind of like an actual picture of a thought.  

 

Even in today's routine QC programs, people graph data because we may see a trend that we can't see in one form but maybe in another.  I think it likely would be the case in analyzing a computer process/function.

 

Ok Edgarcito, you believe that there are many ways to look at information outside of "programmed to produce" an answer.

 

But are these 'many ways' usable by AI?  If you can present evidence that they are, please do so.  Otherwise your belief in these 'many ways' doesn't help you get any closer answering your opening question, does it?   "Do we think AI would conclude there is a God or confirm a particular religion over others?"

 

Yes, there are many ways of looking at information, but if you want AI to answer your question then you've got to do it on the AI's terms, not yours.  And that means looking at information logically.  AI won't work any other way.  Here are three CAPCHA examples of what I mean. CAPCHA = Completely Automated Public Turing test to tell Computers and Humans apart.

 

 

 

 

We humans have no problem passing these tests because we can do what AI cannot.  We can understand the different ways in which the information is presented in these three images.  So, this is a worked example of your 'many ways' to look at information.  Catch is, AI can't look at them and understand them.  And that's how these CAPCHA tests filter out the input of humans from that of bots and automated spam generating programs.

 

So, you see how your 'many ways' belief fails?  If you want AI to answer your question you have to give it what it can understand and work with.  Otherwise, as I've already mentioned, its just Garbage In = Garbage Out.  I'm sorry, but there's just no way round this.

 

 

Thank you,

 

Walter.

 

 

[Edgarcito]

38 minutes ago, walterpthefirst said:

 

So, you see how your 'many ways' belief fails?  If you want AI to answer your question you have to give it what it can understand and work with.  Otherwise, as I've already mentioned, its just Garbage In = Garbage Out.  I'm sorry, but there's just no way round this.

 

 

Thank you,

 

Walter.

 

 

Christ Walter, sometimes I just think you are thick.  I'm not proposing to ask AI anything other than to identify patterns within it's own capacity, for humanity to then compare those findings to religious ideologies.  For example, what if AI's version of light, that "picture of data", array of 1's and 0's, is also found when describing the human retina.  Then one might speculate that that one verse has some weighting for Christianity over another religion that doesn't talk about the light in the eyes.  

 

This is what I am proposing.  Not that difficult....

[Edgarcito]

And besides Walter, are you an AI expert or developer?  How can you say that my idea fails.  I sincerely doubt you can.  Again this is just your ego again that you professed to fight against ruling the day.  Without actually doing the experiment, you have concluded it failing.  Has an equal chance of proving no God as it does a God, yet you can't just have a friendly conversation.  The very reason this place is so fun to come visit....