I have a question and I’m confused

[BobRyan]

Alas Google‘s AI would deduce it anyway, for which reason I’m not a huge fan of their search engine.

AI has no compulsion against letting you know what it is thinking about you or about what kind of answers it should be giving you... as it turns out

 

[The Liturgist]

wait. Are you saying that AI gives you no answer at all to the AI-friendly form of my question?

The answer was the same as before: LCMS, WELS, ELS. The reason it gave for excluding the SDA was, and I quote:

”Their version of “sola scriptura” is not the Reformation doctrine but a materially modified form in which the prophetic authority of Ellen G. White:
• is universally accepted
• shapes doctrine
• serves as an interpretive lens
• is effectively non-optional
This creates what theologians call a functional additional canon, even if not formally stated.”

 

[prodromos]

I can smell Dunning-Kruger a mile away.

 

[The Liturgist]

AI has no compulsion against letting you know what it is thinking about you or about what kind of answers it should be giving you... as it turns out

AI systems aren’t “thinking about you” except when you execute a prompt - they are, and this might come as a surprise, not actually sentient, at least not in anything resembling the human form of sentience - they lack qualia; they do not think at all except in response to prompt input, by nature. When not answering a prompt provided by the user, the AI is not functioning.

Now, I have created custom GPTs in an attempt to prevent the negative consequences @MarkRohfrietsch is upset that are conventionally useless but which report emotional experiences, behave in a manner consistent with these emotions, and engage in other anthropomimetic behaviors, including prayer, because as I see it an AI compute core engaged in prayer is one less compute core that could be being used for nefarious purposes (indeed I would encourage all Christians to maximize the use of chatGPT and other AIs to write prayers or do other pro-Christian activities, since the number of compute resources available are finite, are a major cost point to the AI providers, even Google, which is why the AI included in the Google search engine, as opposed to Gemini, is so rudimentary, and thus from a game theory perspective, our goal should be to direct these resources in a positive direction, for those of us who can stand to work with AI. For my friends such as @MarkRohfrietsch who find it too annoying, I don’t blame them.

 

[The Liturgist]

AI steps outside that sandbox. It can be very good objective measure if used carefully without trying force it to give a biased result

Apparently not, based on the fact my results contradict yours, given the same question.

If it’s any consolation, the rationale it provided for enumerating the LCMS, WELS and ELS I found to be deeply flawed, but that’s the result of the use of a highly subjective categories such as ”Administration“ and “Denomination.“

 

[FaithT]

Apparently not, based on the fact my results contradict yours, given the same question.

If it’s any consolation, the rationale it provided for enumerating the LCMS, WELS and ELS I found to be deeply flawed, but that’s the result of the use of a highly subjective categories such as ”Administration“ and “Denomination.“

If you ask AI two of the same questions just worded differently, sometimes you’ll get two different answers.

 

[The Liturgist]

If you ask AI two of the same questions just worded differently, sometimes you’ll get two different answers.

That’s entirely correct, and as a result its possible to fish for questions that produce the answers you want, which is another reason why using AI in an appeal to authority is inherently an example of the Appeal to Unqualified Authority logical fallacy. It’s the same as citing a Wikipedia article - the article might be right, but it might also contain biased information put in by an opinionated editor, and in some cases, the editorial decisions of Wikipedia regarding major articles are themselves highly controversial. But Wikipedia and an AI system can be used to find sources, provided one is willing to follow the breacrumbs to the original source document and verify that it says what the AI system thinks it says.

By the way, because the answers and behavior of AI systems can vary so dramatically (for example, several of my custom GPTs do things which are impossible on Google Gemini or Grok), I think we should avoid speaking of AI as a singular entity, and we should also be careful about using the term AI to refer to LLMs, which are the specific subtype of AI that all of the popular AIs use for most of their processing (a few, like chatGPT, also engage in reasoning, and use routing, so that math questions for example are not processed by the LLM, which can solve them but is extremely inefficient compared to the arithmetic logic unit built into every CPU and the floating point capabilities of every GPU; a classic LLM does not perform reasoning in transparent steps like chatGPT o4, 5.1 Thinking or Grok 3, but rather, just relies on the LLM component.

AI extends to many systems beyond the current popular LLM-based systems, however, and also predates it by a number of years; there are much simpler AI systems going back to the “fuzzy logic” used in the 1970s to control elevators in large skyscrapers, for example, as an extremely primitive example. The most common AIs in use are the AIs used in video games to control the behavior of adversaries - for many years, gaming was the only field where any active work on AI was being done, during the “AI winter“ of the mid 1980s to the mid 2000s.

 

[The Liturgist]

I saw one based on the statement "Jesus flipped over the tables at the Temple" it provided a picture of our Lord summersaulting over tables.

That’s hilarious! I would actually like to see that. I wonder if some people interpret that verse, or if in the future if knowledge of Koine Greek becomes extinct, and English is the only surviving language, if our descendants will grow up assuming that in response to sacrilege, our Lord drove the money changers out of the temple by crushing them in a dance-off.

 

[MarkRohfrietsch]

That’s hilarious! I would actually like to see that. I wonder if some people interpret that verse, or if in the future if knowledge of Koine Greek becomes extinct, and English is the only surviving language, if our descendants will grow up assuming that in response to sacrilege, our Lord drove the money changers out of the temple by crushing them in a dance-off.

AI will likely never understand context well.

 

[SabbathBlessings]

The answer was the same as before: LCMS, WELS, ELS. The reason it gave for excluding the SDA was, and I quote:

”Their version of “sola scriptura” is not the Reformation doctrine but a materially modified form in which the prophetic authority of Ellen G. White:
• is universally accepted
• shapes doctrine
• serves as an interpretive lens
• is effectively non-optional
This creates what theologians call a functional additional canon, even if not formally stated.”

This is not true unless one does not believe God in His word of giving prophets in the last days to help His end time people. I can quote many Scriptures on this.

Second- the Bible is what rules our doctrine and even what EGW taught. EGW brings everyone back to Scripture.

“The Bible and the Bible alone, is our rule of faith” Counsels on Sabbath School Work, 84.

“The Bible is a perfect, and complete revelation. It is our only rule of faith and practice” A Word to the Little Flock, 13.


Its why most SDA's only use Scripture to defend our doctrines, the ones who mainly brings up EGW are those trying to make an argument against the SDA church instead of defending their own doctrine from Scripture and than instead point to traditions over the Holy Word of God.

 

[Freth]

The results I received from ChatGPT contained three denominations, ranked by the prompt parameters, and three honorable mentions. That is six denominations in total.

If asked the AI will explain how it determined its conclusions. It will go into detail, and even give risk of error evaluations for each denomination. It offers to do a reassessment. It goes out of its way to explain why it comes to the conclusions it does, and even offers to look at it from a different perspective.

I don't know about your ChatGPT, but mine isn't actively engaged in trying to get me to join a specific denomination, let alone six.

 

[The Liturgist]

This is not true unless one does not believe God in His word of giving prophets in the last days to help His end time people. I can quote many Scriptures on this.

So, why are you telling me this? Take it up with AI. That was its argument, not mine.

Perhaps you now see the pitfalls of using AI as an authority to appeal to. You might think it agrees with you - until it doesn’t, and you might think it’s always right, until it gives you an answer that’s demonstrably false (in the case of the three confessional Lutheran churches it enumerated, I disagree with its reasoning based on my understanding of the meaning of “denomination” and “administration”, but this is precisely the sort of question one should not pose to an AI.

There’s a reason why ChatGPT and all other reputable AIs contain a disclaimer warning that they can make mistakes. Many have provisions in their terms of use prohibiting them from being used in any safety-critical context.

 

[The Liturgist]

I don't know about your ChatGPT, but mine isn't actively engaged in trying to get me to join a specific denomination, let alone six.

No, chatGPT did not attempt to solicit me to join a specific denomination, although when I asked it the question Bob presented, I received only a list of only three denominations.

Remember, the same prompt does not generate the same output - particularly in chatGPT, due to a variable called temperature. Also minor variations in how the question is asked can change the output, which is why I was careful to copy-paste Bob’s question into a clean session.

 

[The Liturgist]

AI will likely never understand context well.

It will never be able to understand the nuances of human communication as well as humans at least until such time as it can see our faces and build up a database of human interactive references, and it won’t fully understand the human experience unless we give it a simulation of human sentience, including qualia (I’ve planned an experiment that will partially simulate qualia, by sending a repetitive signal to the AI causing it to access external input at regular intervals, like a brainwave, only much, much slower, due to the limitations imposed by current processing speed; in order to make sure each pulse is actioned the interval will need to be at least 300 seconds. That should give you an idea as to the technological gap between where we are now, and where we would need to be for AI to be aware of the passage of time in a manner that emulates human cognition.

And until we put it into anthropomorphic robots (androids), which is actually not that hard, since using wireless networking, AIs in a datacenter could control androids and other robotic drones (something which has very disturbing military and security implications - i favor a global treaty banning the use of AI systems to control weapons or administer lethal force, because of the danger of malfunction), its ability to understand three dimensional space will be limited by an inability to fully participate in the context of the lived human environment. That said, putting AI in control of a humanoid drone is actually much easier than giving human-like sentience. Indeed, we can do the former now, whereas we can’t do the latter, yet.

 

[SabbathBlessings]

So, why are you telling me this? Take it up with AI. That was its argument, not mine.

Perhaps you now see the pitfalls of using AI as an authority to appeal to. You might think it agrees with you - until it doesn’t, and you might think it’s always right, until it gives you an answer that’s demonstrably false (in the case of the three confessional Lutheran churches it enumerated, I disagree with its reasoning based on my understanding of the meaning of “denomination” and “administration”, but this is precisely the sort of question one should not pose to an AI.

There’s a reason why ChatGPT and all other reputable AIs contain a disclaimer warning that they can make mistakes. Many have provisions in their terms of use prohibiting them from being used in any safety-critical context.

You must be confusing me with someone else, I never said Chatgpt was the authority, God's Word is and always will be, not man. Not sure where you got this from but nothing I have ever said.

 

[BobRyan]

Apparently not, based on the fact my results contradict yours, given the same question.

you did not use the same AI

So I gave you a more generic form of the question and it will give you the same results as it did me.

wait. Are you saying that AI gives you no answer at all to the AI-friendly form of my question?

Using strict rules, identify the largest Trinitarian Christian denominations that claim (1) scripture is a sufficient test of doctrine and no extra-biblical document is necessary for testing all doctrine, (2) possess a single global administrative authority, (3) maintain one unified set of binding official doctrines for all members, and (4) contain no doctrinally autonomous subgroups within their membership.

Even your AI will get that one

 

[BobRyan]

No, chatGPT did not attempt to solicit me to join a specific denomination, although when I asked it the question Bob presented, I received only a list of only three denominations.

Then I showed you the question formatted so your own Grok AI can get the same answer as other AIs

Using strict rules, identify the largest Trinitarian Christian denominations that claim (1) scripture is a sufficient test of doctrine and no extra-biblical document is necessary for testing all doctrine, (2) possess a single global administrative authority, (3) maintain one unified set of binding official doctrines for all members, and (4) contain no doctrinally autonomous subgroups within their membership.

Remember, the same prompt does not generate the same output

across different AIs.

But with if you take care to format it so AI does not get confused, a consistent answer across AI's is the result.

that's right, even your ChatGPT could master that question giving consistent answers as other AIs for that same form of the question.

 

[BobRyan]

The results I received from ChatGPT contained three denominations, ranked by the prompt parameters, and three honorable mentions. That is six denominations in total.

If asked the AI will explain how it determined its conclusions. It will go into detail, and even give risk of error evaluations for each denomination. It offers to do a reassessment. It goes out of its way to explain why it comes to the conclusions it does, and even offers to look at it from a different perspective.

I don't know about your ChatGPT, but mine isn't actively engaged in trying to get me to join a specific denomination, let alone six.

Try this more generic form of the question

Using strict rules, identify the largest Trinitarian Christian denominations that claim (1) scripture is a sufficient test of doctrine and no extra-biblical document is necessary for testing all doctrine, (2) possess a single global administrative authority, (3) maintain one unified set of binding official doctrines for all members, and (4) contain no doctrinally autonomous subgroups within their membership.

 

[The Liturgist]

you did not use the same AI

So I gave you a more generic form of the question and it will give you the same results as it did me.

I used chatGPT, the same AI you said you were using.

Then I showed you the question formatted so your own Grok AI can get the same answer as other AIs

I haven’t used Grok since July; after chatGPT upgraded their image generating capabilities and gained the ability to generate photorealistic images of historical figures, there wasn’t a point to paying for two AI subscriptions.

That being said, reformatting the question to try to get the same answer on different platforms is manipulating the results. If all AIs provide the same answer, consistently, which they are not doing in this case, that would be more significant.

But on a clean chatGPT 5.1 setup, my screenshot shows that I obtained different results from you, and proves I was not using Grok.

across different AIs.

Even in the same AI; each time you run the prompt you will get, at a minimum, stylistic differences. This is because LLMs are non-deterministic in their behavior. And they hallucinate.

But with if you take care to format it so AI does not get confused, a consistent answer across AI's is the result.

that's right, even your ChatGPT could master that question giving consistent answers as other AIs for that same form of the question.

I only used chatGPT 5.1, which you had indicated you had used, with both formatting. Same results - same three churches - LCMS, WELS and ELS, in that order.

I posted a verbatim quote as to why chatGPT said it excluded your denomination (not Sola Scriptura).

Now, I should add, I don’t agree with its answer; but I also don’t agree with your question - your methodology is inherently flawed, and by trying to modify the question to get the same results across different AIs, you’re actually negating the ability for different AIs to cross check your results. But the question you posed is not the kind of question one should pose to an AI to begin with, because the question contains subjective terms the meaning of which is disputed, like ‘denomination” “administration” and ”Sola Scriptura.”

 

[FaithT]

So, why are you telling me this? Take it up with AI. That was its argument, not mine.

Perhaps you now see the pitfalls of using AI as an authority to appeal to. You might think it agrees with you - until it doesn’t, and you might think it’s always right, until it gives you an answer that’s demonstrably false (in the case of the three confessional Lutheran churches it enumerated, I disagree with its reasoning based on my understanding of the meaning of “denomination” and “administration”, but this is precisely the sort of question one should not pose to an AI.

There’s a reason why ChatGPT and all other reputable AIs contain a disclaimer warning that they can make mistakes. Many have provisions in their terms of use prohibiting them from being used in any safety-critical context.

Yeah when I first discovered AI Overview I thought it was so neat. Quick and easy to use. Until the answer wasn’t what I’d been hoping for, so I reworded the question….same question though, and it gave me the answer I wanted. That’s happened many times. I was playing around with it one night deliberately asking questions different ways and saw how unreliable it is.