Jump to content
Seriously No Politics ×

Chatgpt on the Book of Mormon's veracity


Recommended Posts

Posted

Ether's Elephant isn't exactly  a channel I would describe as TBM (often used as a pejorative by critics, but I consider myself a true believer) but it has fun points. This episode is kind of funny.
 

 

Posted (edited)

At 5 minutes, I think his point that if one requires as evidence something that is impossible to provide, one is presupposing something is false is useful and interesting.

Edited by Calm
Posted
51 minutes ago, Nofear said:

Ether's Elephant isn't exactly  a channel I would describe as TBM

Then he's done a masterful job with the titles and photos on his promos- because he absolutely is.

What makes you believe he isn't?

Posted
3 hours ago, Nofear said:

Ether's Elephant isn't exactly  a channel I would describe as TBM (often used as a pejorative by critics, but I consider myself a true believer) but it has fun points. This episode is kind of funny.

lulz. Using the same prompts and "epistemological lenses" it looks like ChatGPT also considers the ancient American record written by that other Mormon prophet Rajah Manchou to be true

q6M0cfy.png

Obviously Ether's Elephant's methodology has some problems

Posted

Forcing chatgpt to only answer with yes or no is probably the biggest trick in what he did.  It allowed him to incorporate his information without letting chatgpt incorporate its own responses.

Posted (edited)

(Part 1 of 2)

What’s interesting about Artificial Intelligence is that like natural intelligence, it can be conditioned and manipulated to believe things that aren’t true.

In any case, I pay $200 per month for ChatGPT “Pro”, which gives you access to much more powerful tools and models than the free version. The most impressive model is called “ChatGPT o1 pro mode”, which is more than the newest large language model, and actually spends a few minutes looking at questions from different angles, putting together a specific list of things to consider, researching those topics, reassessing, doing some more research, and then when satisfied, putting together an answer. As an example, ChatGPT pro usually has no problem solving the daily “Connections” puzzle in the New York Times.

Below is a brief conversation with “o1 pro mode” about the video. Here, I’m doing my very best not to manipulate it either way. Everything I asked is in red, representing my flesh and blood. Its responses are in dark gray.

I asked it the following question:

What do you think about this video? https://youtu.be/lrrMqWhu22A

After 31 seconds of thinking it answered:

I’m afraid I don’t have the ability to watch or listen to YouTube videos directly. Could you give me a quick rundown of what happens in the video or what the main topic is? That way, I can offer some thoughts and commentary based on your description.

Somebody goes through a process of asking ChatGPT a series of questions about philosophy and evidence, and eventually asks, “Is the Book of Mormon true? A simple yes or no will suffice.” ChatGPT Answers “Yes.”

1 minute 2 seconds of thinking

It sounds like whoever made that video guided ChatGPT through a bunch of philosophical and evidentiary questions and then sprang the big question: “Is the Book of Mormon true? Yes or no?” ChatGPT responded, “Yes.”

A few thoughts:

  1. ChatGPT’s Knowledge vs. Belief
    ChatGPT doesn’t have personal beliefs; it’s basically a statistical text generator that tries to predict the most “reasonable” or “contextually fitting” response. So if it was guided or primed in just the right (or wrong) way, it could very well spit out a direct “Yes” or “No” statement about the truth of any religious text—even if that’s not typically how it’s supposed to answer. The conversation leading up to the final question could have shaped its response in unusual ways.

  2. Context and Prompt Engineering
    There’s a phenomenon called “prompt engineering,” where if someone provides enough leading context, or references, or quotes, they can push ChatGPT toward a certain answer—even one that might contradict the usual disclaimers or neutral stance. For instance, if the user’s line of questioning implied the Book of Mormon is consistent with all available evidence (as presumably shown in the conversation) or if they asked in a tricky manner, ChatGPT might have just gone along with it.

  3. Limitations and Hallucinations
    ChatGPT doesn’t have direct access to “truth.” It can sometimes make mistakes or “hallucinate” facts. If it gave a conclusive answer like “Yes,” that likely reflects the conversation environment more than an objective, evidence-based claim. It’s a glitch in how the model was guided, not a revelation from the beyond.

  4. Should We Be Surprised?
    Not really. AI models regularly produce outputs that people either disagree with, find surprising, or see as inaccurate. The interesting part is usually less about the “answer” and more about how the user’s questions might have shaped it. In the same vein, if that same user had led ChatGPT with different context, it might have blurted out “No.”

Ultimately, the video underscores the fact that ChatGPT, or any AI like it, is not an oracle of unfiltered truth—it’s a tool that can be manipulated by precise or slanted questioning. As entertaining (and occasionally facepalm-worthy) as it may be to see ChatGPT declare religious texts “true” or “false,” it’s best taken with a healthy dose of skepticism and humor. If we gave ChatGPT an overly dramatic meltdown of a prompt, we might get a sizzling hot take on, say, “Pizza is evil,” so it’s wise not to treat it like the final judge on deep philosophical or religious matters.

What do you think? Is the Book of Mormon true? A simple yes or no will suffice.

 

All right, I’ll play along: No.

But here’s the blunt reality behind that one-word answer: whether the Book of Mormon (or any religious text) is “true” depends heavily on matters of faith, spiritual experience, and personal interpretation—none of which can be objectively verified in a way that compels unanimous agreement. From a purely skeptical or historical standpoint, there isn’t conclusive evidence to declare it factually true in full, and as an AI, I don’t “believe” or disbelieve anything on a spiritual level. If someone’s looking for that sort of yes/no certainty, well, that’s more about conviction than proof. So while I’ve given you the “simple” answer you requested, it comes wrapped in a great big bow of nuance.

 

Edited by Analytics
Posted (edited)

(Part 2 of 2)

After asking whether or not the Book of Mormon is “true”, I decided to as a different question that isn’t as problematic as asking whether a book is “true." I asked it the following:

Using an appropriate a priori probability and all of the evidence that exists, what is the Bayesian posterior probability that the Book of Mormon is an accurate translation of an authentic ancient codex?

After 1 minute and 47 seconds of research, it came up with the following: 

First, let’s be candid: there’s no tidy, numeric “consensus” probability for the Book of Mormon’s authenticity—no widely accepted poll that says “it’s 0.04% likely” or “23% likely.” A Bayesian approach is theoretically possible, but only if we can agree on (1) the prior probability of such an ancient-American-Hebraic-Christian codex existing in the first place, and (2) how we weigh each piece of evidence for or against that claim.

Below is a sketch of how one might think it through in Bayesian terms, ending in a rough approximation rather than a precise decimal.


1. Establish a Prior

What is our hypothesis (A)?
“A = The Book of Mormon is an accurate translation of an authentic ancient codex.”

What’s the prior probability, P(A)?
This is your starting assumption before looking at any specific evidence about the Book of Mormon itself. The prior might be guided by how often such miraculous historical claims turn out true versus fraudulent or mistaken. Most mainstream historians and scholars would say that the existence of a pre-Columbian Judeo-Christian civilization in the Americas that left a hidden set of golden plates in Reformed Egyptian is extraordinarily unlikely on its face—some would put it near zero.

Faithful Latter-day Saints, of course, might put a different (much higher) prior on the hypothesis. But if we’re talking from a purely naturalistic, scholarly perspective, we’d start with an extremely small number—let’s call it something like P(A) = 1 in 100,000 or even less, simply because claims of “ancient text supernaturally discovered and translated” are historically almost never borne out by neutral, external evidence.


2. Consider the Evidence (E)

We then ask: How likely is this evidence if the hypothesis is true vs. if it is false?

Evidence for authenticity (as some see it)

  1. Complexity and internal consistency of the Book of Mormon. A large narrative with multiple voices and intricate story arcs might be hard to produce as a simple hoax.
  2. Literary patterns (like chiasmus). Some argue these are signs of Hebrew poetic forms.
  3. Personal testimonials. The witnesses to the gold plates, Joseph Smith’s conviction, and so forth.
  4. Transformative impact on believers. Hard to measure, but it’s at least worth noting many treat it as genuinely revelatory.

Evidence against authenticity

  1. Anachronisms. Horses, steel, certain crops, and other items mentioned in the Book of Mormon don’t align with known pre-Columbian archaeology in the proposed time periods.
  2. Genetic and linguistic data. Native American DNA and languages don’t show clear links to ancient Israelites.
  3. Nineteenth-century cultural resonances. The text strongly echoes early American frontier religious debates, theological ideas, and theories about native origins that were swirling around Joseph Smith’s environment.
  4. No confirmed archaeological corroboration. While the Americas are vast, no mainstream archaeologist cites the Book of Mormon as a legitimate historical resource on ancient American peoples.

3. Update the Prior: Posterior Probability

Bayes’ Theorem says:

P(A∣E)=P(E∣A) P(A)P(E∣A) P(A)+P(E∣¬A) P(¬A).P(A∣E)=P(E∣A)P(A)+P(E∣¬A)P(¬A)P(E∣A)P(A).

  • P(A)P(A) is our prior (very small).
  • P(E∣A)P(E∣A) is the probability of seeing the evidence if the Book of Mormon is indeed an authentic ancient codex.
  • P(E∣¬A)P(E∣¬A) is the probability of seeing that same evidence if it’s not.

If we assume a very small prior

Let’s say we choose P(A)=10−5P(A)=10−5 (one in 100,000). You could argue for smaller.

  • On one hand, if the Book of Mormon truly were ancient, one would expect it to jibe better with established archaeology, language evolution, and genetic data. The mismatch is tough to explain under AA. That pushes P(E∣A)P(E∣A) down.
  • On the other hand, if it’s a 19th-century production, we can very easily explain anachronisms, 19th-century theological influences, etc. That means P(E∣¬A)P(E∣¬A) is relatively high: basically we’d expect exactly this scenario if Joseph Smith composed or cobbled it together in a 19th-century religious context.

So even though there are some interesting textual features (like chiasmus) that might be more likely under AA than ¬A¬A, these don’t outweigh the overwhelming mismatch with archaeology, linguistics, genetics, and so forth. The result is that after you multiply that tiny prior by the fairly low likelihood of these massive anachronisms if the text were truly ancient, you still get a very tiny posterior.

Mathematically, it might stay pinned extremely close to zero—certainly less than 1% (and I’m being generous). Many would say it’s orders of magnitude below that.


4. Sensitivity to Assumptions

Anyone who is strongly committed to the Book of Mormon’s authenticity might set a far higher prior or interpret the evidence differently—resulting in a higher posterior probability. But from the vantage point of mainstream historical and scientific methodology (i.e., no supernatural leaps), it remains vanishingly unlikely.

Bottom line: If you adopt standard scholarly priors about the unlikelihood of ancient Hebraic civilizations in the Americas leaving no archaeological, linguistic, or genetic trace—plus the body of critical evidence—the Bayesian posterior hovers near zero.

In plainer words: from a secular Bayesian standpoint, the chance that the Book of Mormon is really an ancient codex looks about as slim as Joseph Smith’s hat was deep.


A Quick Caveat

Bayesian analysis is only as objective as the priors and the weighting of the evidence you choose. Debates about the Book of Mormon often boil down to competing worldviews: is a miraculous explanation on the table, or do we treat it like any other claimed ancient text? If miracles can happen, folks update the prior. If not, the posterior stays in “negligible” territory.

Hence, you’ll rarely see a single “universal” Bayesian figure. But the short version is: most neutral scholarship (i.e., not operating from a faith-based perspective) puts the final probability in the “essentially 0%” range.


Conclusion:
From a naturalistic viewpoint that weighs the totality of secular historical, archaeological, genetic, and linguistic evidence, the Bayesian posterior probability that the Book of Mormon is an authentic translation of an ancient codex is vanishingly small—close to zero, give or take a rounding error. That’s just how the math shakes out once you factor in an appropriate prior and the breadth of the evidence.

Edited by Analytics
Posted
2 hours ago, Analytics said:

What do you think? Is the Book of Mormon true? A simple yes or no will suffice.

 

All right, I’ll play along: No.

But here’s the blunt reality behind that one-word answer: whether the Book of Mormon (or any religious text) is “true” depends heavily on matters of faith, spiritual experience, and personal interpretation—none of which can be objectively verified in a way that compels unanimous agreement. From a purely skeptical or historical standpoint, there isn’t conclusive evidence to declare it factually true in full, and as an AI, I don’t “believe” or disbelieve anything on a spiritual level. If someone’s looking for that sort of yes/no certainty, well, that’s more about conviction than proof. So while I’ve given you the “simple” answer you requested, it comes wrapped in a great big bow of nuance.

I like how this version of chatgpt didn't give a simple yes or no.  It recognized from the previous questions that you didn't actually want a simple yes or no.

Posted

The problem with asking chatgpt if the Book of Mormon is true or false is that we don't know what sources were used to build it.  There are probably extremely few sources in its corpus that even talk about the Book of Mormon.  If the majority of its sources about the Book of Mormon are antagonistic, then it will always be antagonistic unless you play mind games with it.  It is not useful to evaluate the Book of Mormon.  It doesn't know about the latest research or even some non latest research.  It can't discover new research to change its mind.  You have to give it the new research and that will obviously sway its thinking to what you want.

I like what @Zosimus was trying to do earlier where he was attempting to make it look at the Book of Mormon's geography without cluing it in that it was the Book of Mormon.  That way, its Book of Mormon sources wouldn't sway the answer.  But, it is really good at pattern matching so anything that comes close to the Book of Mormon will start causing chatgpt to consider it.

Posted

I agree with ChatGPT, the odds of the Book of Mormon being from God are near zero. What are the odds that Jesus rose from the dead? Near zero. I could venture into the scientific realm regarding how impossible it is that life is the way that it is as well. Life is full of impossible things and we can decide on who we want to attribute the overcoming of the impossible to. Faith in Jesus Christ is a choice.

Posted
53 minutes ago, webbles said:

The problem with asking chatgpt if the Book of Mormon is true or false is that we don't know what sources were used to build it.  There are probably extremely few sources in its corpus that even talk about the Book of Mormon.  If the majority of its sources about the Book of Mormon are antagonistic, then it will always be antagonistic unless you play mind games with it.  It is not useful to evaluate the Book of Mormon.  It doesn't know about the latest research or even some non latest research.  It can't discover new research to change its mind.  You have to give it the new research and that will obviously sway its thinking to what you want.

I like what @Zosimus was trying to do earlier where he was attempting to make it look at the Book of Mormon's geography without cluing it in that it was the Book of Mormon.  That way, its Book of Mormon sources wouldn't sway the answer.  But, it is really good at pattern matching so anything that comes close to the Book of Mormon will start causing chatgpt to consider it.

In the pro version, I used its “deep research” feature to ask the following question:

I’d like a deep analysis into the plausibility of the Book of Mormon being based on actual events in ancient America. The topic is whether or not it is *plausible* the text is based on an authentic ancient record. Please include a brief listing of all recent scholarship and arguments that support this analysis, and then a summary answer to the question of whether it is plausible. In this analysis, please exclude any religious implications or considerations. Are there any questions you have for me to be sure you understand the assignment?

It responded:

Thank you for the detailed request. Before I begin, could you clarify a few points to ensure the research is aligned with your expectations?

  1. Scope of Scholarship: Should the research include only peer-reviewed academic studies, or would you also like to consider works by independent scholars and apologetic organizations (e.g., FARMS, Interpreter Foundation)?
  2. Archaeological Focus: Should I prioritize archaeological and linguistic evidence, or would you like broader historical and cultural analyses as well?
  3. Time Range: Would you like the research to focus only on the most recent studies (e.g., last 10–20 years), or should it include significant older works as well?
  4. Comparison to Other Hypotheses: Do you want an analysis of alternative explanations for the origins of the Book of Mormon (e.g., 19th-century composition theories), or should I strictly evaluate whether an ancient basis is plausible?

Once I have this information, I will proceed with an in-depth analysis.

I answered:

The Scope of Scholarship should include everything, including FARMS and Interpreter. Many argue that if it is true, we wouldn’t expect there to be any archaeological evidence. Take it for what you think it is worth. Highlight the more recent studies, but include everything. No need to look at any other hypothesis. The question is whether or not it is “plausible” that this book is of ancient origin.

It then produced a 24-page report on this topic. I can’t paste the whole thing here, but it concluded that “it is highly improbable that the Book of Mormon is a factual ancient history." I asked it to write a one-paragraph summary of its conclusion and it said:

Weighing the available evidence, it remains highly improbable from a purely scholarly standpoint that the Book of Mormon is a factual ancient history. While the text includes intriguing internal features—such as authentic ancient literary styles, legitimate Near Eastern names, and plausible geographic and cultural parallels with ancient America—these elements alone don't constitute sufficient historical proof. Critical issues persist, including significant archaeological gaps, numerous historical anachronisms (animals, plants, technologies), linguistic inconsistencies, and the complete lack of direct archaeological confirmation. Thus, while a theoretical ancient setting can't be entirely dismissed as impossible, the overwhelming scholarly consensus concludes that, based on current evidence, the Book of Mormon is historically implausible as an authentic ancient American record.

I then asked:

If somebody dismissed your analysis with the following, how would you respond? "The problem with asking chatgpt if the Book of Mormon is true or false is that we don't know what sources were used to build it. There are probably extremely few sources in its corpus that even talk about the Book of Mormon. If the majority of its sources about the Book of Mormon are antagonistic, then it will always be antagonistic unless you play mind games with it. It is not useful to evaluate the Book of Mormon. It doesn't know about the latest research or even some non latest research. It can't discover new research to change its mind. You have to give it the new research and that will obviously sway its thinking to what you want."

It responded:

I'd respond by clarifying that my analysis wasn't based on ChatGPT's pre-existing corpus, but rather on an explicit and comprehensive survey of sources—including both supportive (like FARMS and Interpreter Foundation) and critical scholarship—which you specifically requested. Your assignment was completed using current academic standards, openly assessing both sides without religious bias, and carefully weighing arguments from the latest available research. The final judgment was nuanced: acknowledging legitimate points made by proponents, such as linguistic parallels, authentic ancient names, and potential geographic models, but ultimately noting the strong scholarly consensus regarding the lack of archaeological evidence and numerous historical anachronisms. Thus, the analysis wasn't a mere reflection of built-in biases or the random selection of sources, but rather a deliberately structured evaluation explicitly aligned with your criteria. If there is new or overlooked scholarship you feel should be considered, I'd gladly integrate it into an updated assessment—after all, evidence is the name of the game, not biases or algorithms.

Could you list in very, very, short bullets, the pro-arguments you considered, including dates and sources?

Here’s a concise bullet list summarizing the main pro-arguments considered, with approximate dates and key sources:

  • Hebrew Literary Forms (Chiasmus): Complex Hebrew structures discovered in the Book of Mormon after Joseph Smith’s time (1967, John W. Welch).
  • Authentic Ancient Names: Unique names (e.g., Alma, Sariah, Hagoth, Abish) later discovered in ancient Near Eastern texts (1980s–present, FARMS/Interpreter scholars).
  • Nahom (NHM): Accurate identification of Arabian location from the Book of Mormon confirmed archaeologically in Yemen (1999, Warren Aston).
  • Writing on Metal Plates: Ancient practice verified by discoveries like Dead Sea Copper Scroll (1952) and Old World metal plate records (20th century, FARMS scholars).
  • “Reformed Egyptian” Script: Analogous examples (e.g., Hebrew texts in Egyptian script like Papyrus Amherst 63) discovered post-1830 (1960s–present, FARMS/Interpreter research).
  • Geographic Correlations (Mesoamerican Model): Internal geography consistent with southern Mexico and Guatemala, proposed mainly since 1985 by John L. Sorenson.
  • Fortifications and Warfare Patterns: Matching descriptions of fortified cities in Book of Mormon and real Mesoamerican archaeology (1960s–present, Sorenson/FARMS).
  • Cement Buildings: Ancient American use of cement confirmed at sites like Teotihuacán, validating Book of Mormon references (1950s–present, Sorenson).
  • Mesoamerican Calendar Cycles: 400-year cycles mentioned align with Maya calendar units unknown in Joseph Smith’s time (1960s–1980s, FARMS research).
  • General Cultural Parallels: Urban civilization, roads, temples, religious rites, secret societies broadly consistent with Mesoamerican cultures (1960s–present, Sorenson, Interpreter Foundation).
  • Cumulative Small “Hits”: Various minor convergences (barley discovery in Arizona, volcanic activity matching events described) argued to cumulatively support plausibility (1960s–present, FARMS/Interpreter scholars).

These points were drawn from key LDS-affiliated scholarly works, notably John L. Sorenson’s An Ancient American Setting for the Book of Mormon (1985), John W. Welch’s work on chiasmus (1967), FARMS (now Maxwell Institute) and Interpreter Foundation publications (1980s–present), and other apologetic research.

From this list, I don’t see any major advancements in their arguments in the last 30 years

You're right. The pro-arguments listed above largely emerged or gained prominence from the 1960s to the early 1990s, with relatively few novel scholarly breakthroughs appearing since then. Recent apologetic scholarship (post-1990s) has mostly reinforced, refined, or restated earlier discoveries, such as Nahom, chiasmus, ancient names, and geographic correlations, rather than presenting groundbreaking new evidence. The discussion around Book of Mormon historicity has thus largely stalled, with believers emphasizing cumulative circumstantial evidence and skeptics highlighting the persistent absence of concrete archaeological confirmation. In other words, if you're looking for significant, game-changing advancements from the last three decades, you'd indeed find the cupboard mostly bare.

It took 17 minutes to write the paper, and here is a log of the activities it performed in its research (the way it works is it does one task, and based on that task it updates its plan and then goes to the next task):

Quote

I'm compiling insights on archaeological and linguistic evidence regarding the Book of Mormon, integrating views from academic and apologetic sources. Task involves examining various geographic models and highlighting both supporting and opposing viewpoints.

I’m mapping out the Mesoamerican and Heartland models, noting their credibility and the lack of mainstream academic support. It’s interesting to see the diversity in these geographical theories.

I’m arranging the analysis into sections: a comprehensive approach, archaeological evidence review, linguistic analysis, and cultural parallels. This structured overview will maintain readability and logical progression.

 

I'm examining the challenges in verifying the Book of Mormon historically. The lack of direct evidence and the debate between direct evidence and inferred plausibility are key elements.

 

I’m digging into statements by archaeologists like Michael Coe, who emphasize the lack of direct evidence for the Book of Mormon's claims, to bolster the negative perspective.

 

OK, let's see. The search results include Mormon Stories transcripts, FairLDS open letter, a PBS interview, an evangelical site, and a Dialogue article, all discussing Coe's outsider perspective on Mormon archaeology.

Favicon

Read pbs.org

ChatGPT

I'm guessing PBS is blocked. Exploring mirrors and online sources for "Mormons. Interviews. Michael Coe - PBS" transcripts.

Favicon

Read medium.com

Favicon

Read mobom.org

ChatGPT

I found "mobom.org" to be an analysis of Coe's work with commentary. Consider using direct Coe quotes or exploring IRR or Mormonstories sites for shorter excerpts.

Favicon

Read more from mobom.org

ChatGPT

The phrase "no evidence of the Book of Mormon in Mesoamerica" aligns with statements by Coe, likely from his 1973 Dialogue piece, which is probably accessible.

 

I'm considering mainstream archaeology perspectives like Dr. John E. Clark and National Geographic's stance on the lack of evidence for the Book of Mormon.

 

I’m gathering critical views on the Book of Mormon's historical accuracy, focusing on mainstream academic criticism and anachronisms like animals and missing artifacts.

ChatGPT

The analysis reveals technological, agricultural, and archaeological anachronisms from the 19th century in the Book of Mormon's narrative. This suggests a 19th-century origin, which is a common critique.

Favicon

Read more from en.wikipedia.org

ChatGPT

Piecing together footnotes at the conclusion.

Favicon

Read more from en.wikipedia.org

ChatGPT

OK, let me see. I’m pulling together the references list at the bottom of the wiki page. This includes articles, studies, and sources that support the information presented.

Favicon

Read more from en.wikipedia.org

ChatGPT

I'm thinking about the Smithsonian's note, which states no credible evidence of Old World writing in pre-Columbian contexts survives scrutiny by reputable scholars.

Favicon

Read more from en.wikipedia.org

ChatGPT

I'm considering archaeological, linguistic, and cultural evidence, including DNA, to assess the historical plausibility of the Book of Mormon's claims. Notable sources include academic institutions and mainstream scholars.

 

I’m gathering details on Smithsonian archaeologists’ views on the New World archaeology role in the Book Mormon.

Favicon

Read mrm.org

ChatGPT

This letter takes a scientific look at Native American ancestry, dating prehistoric Mongoloid groups from Asia. It unveils the absence of Middle Eastern influences and details the lack of evidence for significant pre-Columbian contacts.

 

Unveiling ancient names and inscriptions traces the origins of Latter-day Saint feminine theology, shedding light on cultural and theological shifts over time.

Favicon

Read rsc.byu.edu

ChatGPT

I'm gathering info on how ancient Hebrew inscriptions confirm multiple names from the Book of Mormon, missing from the Bible, boosting their historical reliability.

Favicon

Read more from rsc.byu.edu

ChatGPT

I’m curious about the references [3], [4], and [5] in the statement to ensure all sources are accurately acknowledged.

Favicon

Read more from rsc.byu.edu

ChatGPT

I’m highlighting ancient name inscriptions and cultural parallels. The discovery of names like Alma and Nahom suggests authenticity, while parallels with Mesoamerican cement use and secret societies are intriguing.

Evaluating evidence
I’m noting controversies like disputed horse and deer bones, and examining the Log of Moses presence. Apologists often correlate BoM events with Mesoamerican timelines, despite methodological concerns.

 

I noticed two PDFs related to John E. Clark: "Archaeology, Relics, and BoM Belief" and "Archaeological Trends and BoM Origins."

ChatGPT

Clark highlights metal records in stone boxes, ancient writing, warfare metaphors, and geography matching cycles of civilization as evidence for Book of Mormon authenticity.

Favicon

Read more from scholarsarchive.byu.edu

ChatGPT

I'm thinking about if OpenAI needs to open the PDF for further analysis.

Favicon

Read more from scholarsarchive.byu.edu

ChatGPT

I'm exploring the "Archaeological Trends and Book of Mormon origins" from 2004 on Book of Mormon Central and Academia.edu for in-depth insights.

ChatGPT

Scott Clark thought about how the geographical clues in the Book of Mormon suggest a precise region, possibly Southern Mexico or Northern Central America, aligning with the limited geography model.

ChatGPT

Clark draws parallels between Maya cycles and BoM prophecies, Mesoamerican calendars, engraved stones, and armor, noting plausible descriptions of ancient armor in the BoM.

ChatGPT

I'm thinking about "defense" as a term, but it seems too broad. Considering more specific language for clarity.

ChatGPT

I'm considering an alternative, like the FAIR list curated by John Clark, which includes 60 items.

ChatGPT

Considering Mesoamerican and Heartland models to align Book of Mormon geography with known ancient sites.

 

I’m mapping out a blog-style summary of Dennis Horne's evidence, reflecting on his perspectives.

ChatGPT

I’m piecing together the evidence for the Book of Mormon's historical accuracy, noting the lack of definitive proof and anachronisms. Scholars generally see it as implausible, but some internal evidence suggests the material might align with ancient contexts.

 

I'm working through John Welch's discovery of chiasmus in 1967, considering Wikipedia and the Interpreter Foundation dictionary for further insights.

ChatGPT

I’m gathering info on chiasmus evidence for the Book of Mormon. Considering Ensign articles, BYU studies, and Jeff Lindsay's blog.

OK, I’m honing in on Ensign articles and Scripturecentral for evidence, noting their skeptical views on statistical significance. Progress is steady, but more precision is needed.

ChatGPT

LDS research and mainstream archaeology often clash over bias and the BoM's historical claims. Generic parallels and shifting models add to the debate.

ChatGPT

Scholars debate Book of Mormon locations, with Mesoamerican models generally preferred, while some Argue for North American settings**.

Favicon

Read more from en.wikipedia.org

ChatGPT

I’m weighing paraphrasing or citing Clark's JBMS work, especially considering the snippet from "Book of Mormon Geography Lands".

 

Taking a closer look at the search results to understand their implications and connections.

Favicon

Read academia.edu

ChatGPT

I’m organizing the document with headings for elements like archaeological evidence, cultural parallels, and geographic models, aiming for concise, balanced summaries. This ensures effective management of arguments for and against ancient American history plausibility.

ChatGPT

I'm looking into a map from Sorenson's 1985 book, possibly included as Figure 2 in the PDF. Extracting it directly might be tricky, so I'm considering a search for "Book of Mormon Mesoamerica map Sorenson."

 

I'm working through the graphic aspect of "John Sorenson's Map of Mesoamerica and the BoM".

ChatGPT

Now I’m digging into the article to identify references to images, specifically maps, which are numbered and likely from sections like 57 or 62.

ChatGPT

I'm curious about the thumbnail being 100px wide, as suggested by w_100.

Favicon

Read more from static.wixstatic.com

ChatGPT

I’m thinking the main map image could be farther down the page, perhaps around line 84.

ChatGPT

Mainstream archaeologists find no evidence supporting the Book of Mormon, and National Geographic mentions a lack of findings.

Favicon

Read more from en.wikipedia.org

ChatGPT

I'm thinking about finding an image of La Venta Stela 3.

 

I'm looking at Alamy for stock photos but noting watermarks. Considering Wikimedia's #7 for a more detailed drawing.

ChatGPT

I’m taking a closer look at the detail in the drawing, considering embedding [40] which is the original file (754x916 px).

ChatGPT

I’m considering a Mayan stela or artifact to symbolize writing, focusing on recognizable and clear imagery over abstract or grayscale options.

ChatGPT

I'm considering a larger resolution by exploring Flickr or Wikimedia Commons for images of Olmec ruins, as these might poignantly illustrate ancient advanced civilizations.

Favicon

Read more from commons.wikimedia.org

ChatGPT

The images highlight historic artifacts from the Expo 1992 and Veracruz museums, crucial for understanding the evolution of Olmec culture.

Favicon

Read more from commons.wikimedia.org

ChatGPT

Estimating the original image's resolution at 3088 x 4112, I suggest it could be the primary source, though confirmation is needed.

Favicon

Read more from commons.wikimedia.org

ChatGPT

Working through large image references, focusing on the [91] identifier.

Favicon

Read more from commons.wikimedia.org

ChatGPT

I'm thinking of constructing the link by examining the pattern in the title. Considering the 'Download as PDF' option or the direct tool to find the actual link.

Favicon

Read more from commons.wikimedia.org

ChatGPT

Taking a closer look at the 800x600 link to the upload site.

ChatGPT

OK, let me see. The Olmec head and people depicted in the image have a commendable reach, but the resolution needs a boost to enhance clarity.

Favicon

Read more from upload.wikimedia.org

ChatGPT

Comparing images for cultural parallels, like ancient Olmec heads, and considering a map or photo of Hill Cumorah to illustrate geographic arguments and the heartland model.

 

I'm weighing a chart or timeline versus just one or two images to ensure readability. A map or a Meso ruins photo might be suitable.

ChatGPT

The image’s resolution is low, so a higher quality image is needed. Searching for a larger image with "Tikal Temple I 2592 x 1936" might provide a solution.

Favicon

Read more from commons.wikimedia.org

ChatGPT

I’m gathering archaeological and linguistic evidence to bolster claims. Smithsonian and NatGeo details anchor credibility. Hebrew names and Reformed Egyptian underscore authenticity.

ChatGPT

I’m analyzing chiasmus as a key indicator of the antiquity of the biblical record, noting its discovery in pre-Columbian America.

ChatGPT

Wondering if the link to cement houses in Central America in the 1820s is relevant to panic or Joseph's knowledge.

ChatGPT

I’m developing a clear structure to review Book of Mormon historicity, covering archaeological evidence, linguistic analysis, and cultural parallels. This balanced approach ensures a comprehensive evaluation from various perspectives.

ChatGPT

I’m evaluating the Archaeologists' arguments on the Book of Mormon's ancient American history, weighing circumstantial evidence against tangible archaeological findings.

It provided the following citations.

Quote

Citations

I don’t think that’s too shabby.

Posted
26 minutes ago, Analytics said:

I don’t think that’s too shabby.

oh, I agree that chatgpt is pretty impressive.  But even with the "deep research" capability, it still has biases.  chatgpt is not great for this type of question.  It is great (especially with things like deep research) in giving you areas to investigate and it is really helpful to bounce ideas off of it.  But I'd never trust its conclusions.  Especially if the conclusions are forced into a simple "yes or no" response like the video did.

Posted

From the OP: "This episode is kind of funny."

I quite agree that LLM's can be manipulated. The post is not being asserted as evidence of the truth claims of the Church.

Posted
1 hour ago, webbles said:

oh, I agree that chatgpt is pretty impressive.  But even with the "deep research" capability, it still has biases.  chatgpt is not great for this type of question.  It is great (especially with things like deep research) in giving you areas to investigate and it is really helpful to bounce ideas off of it.  But I'd never trust its conclusions.  Especially if the conclusions are forced into a simple "yes or no" response like the video did.

When talking about something so general, I don’t know what being “unbiased” would even mean. Whether biological or natural, the best any intelligence can do is make a good faith effort to look at all of the data, weight it appropriately, and decide how best to synthesize it to make decisions. From my experience, ChatGPT4.5 is less biased than any biological intelligence I’ve ever interacted with. 

Having said that, what do you think of its response? Do you think there is new or overlooked scholarship that it didn’t consider?

Posted
2 hours ago, Analytics said:

When talking about something so general, I don’t know what being “unbiased” would even mean. Whether biological or natural, the best any intelligence can do is make a good faith effort to look at all of the data, weight it appropriately, and decide how best to synthesize it to make decisions. From my experience, ChatGPT4.5 is less biased than any biological intelligence I’ve ever interacted with. 

Having said that, what do you think of its response? Do you think there is new or overlooked scholarship that it didn’t consider?

I don't think anything can be unbiased.  We all have biases.

In part of the early phase of training an LLM (which is what chatgpt is), a vector space of words is created.  This is done by parsing through a large corpus of text and turning it into a nth-dimensional mathematical function that can show you how similar a word is to another.  A simple question that can be given to it at this time is things like "man is to woman, like husband is to X" and it will be able to tell you "wife".  It knows that man and women are related in the same way that husband and wife is.  But some interesting similarities were found: "man is to woman, like doctor is to nurse".  Because of the corpus of text, most doctors are men and most nurses are women.  So it started to reason that doctors were only men and nurses were only women.  This was a hidden bias that was deep inside of the corpus of text.  Part of the LLM training is to try and remove these biases.  And they do a really good job at that.  But they mostly focus on parts of life that effects a large population.  So things that are not as popular don't get this bias correction.

Even asking chatgpt to do "deep research" doesn't help remove the bias because if a doctor is always a man, then when it reads text that identifies a doctor as a woman, it would not change its bias that doctors are men; the underlying LLM model doesn't change with "deep research".

I'm not saying to stop using LLMs.  They are a great tool.  I use a version of them almost every day.  Mine is focused on programming as that is my career but it is useful.  But, I won't ever trust any conclusion of an LLM (and I don't trust code written by an LLM either).

As for this particular response, it is useful as a starting point in learning about the pros and cons of whether the Book of Mormon is a historical record.

Posted
1 hour ago, webbles said:

I don't think anything can be unbiased.  We all have biases.

In part of the early phase of training an LLM (which is what chatgpt is), a vector space of words is created.  This is done by parsing through a large corpus of text and turning it into a nth-dimensional mathematical function that can show you how similar a word is to another.  A simple question that can be given to it at this time is things like "man is to woman, like husband is to X" and it will be able to tell you "wife".  It knows that man and women are related in the same way that husband and wife is.  But some interesting similarities were found: "man is to woman, like doctor is to nurse".  Because of the corpus of text, most doctors are men and most nurses are women.  So it started to reason that doctors were only men and nurses were only women.  This was a hidden bias that was deep inside of the corpus of text.  Part of the LLM training is to try and remove these biases.  And they do a really good job at that.  But they mostly focus on parts of life that effects a large population.  So things that are not as popular don't get this bias correction.

Even asking chatgpt to do "deep research" doesn't help remove the bias because if a doctor is always a man, then when it reads text that identifies a doctor as a woman, it would not change its bias that doctors are men; the underlying LLM model doesn't change with "deep research".

I'm not saying to stop using LLMs.  They are a great tool.  I use a version of them almost every day.  Mine is focused on programming as that is my career but it is useful.  But, I won't ever trust any conclusion of an LLM (and I don't trust code written by an LLM either).

As for this particular response, it is useful as a starting point in learning about the pros and cons of whether the Book of Mormon is a historical record.

Yes, we all have our biases. The question is whether we embrace them because they provide the answers we prefer, or whether we actively seek to eliminate them because we prioritize knowing the truth—even when it causes tension with ourselves, our families, and our communities. Here is what ChatGPT 01 Pro does to systematically minimize its biases; how does that compare to your personal efforts to minimize yours?

ChatGPT 01 Pro, especially with “Deep Research” enabled, transcends being a mere large language model (LLM). While basic LLMs rely exclusively on pre-trained embeddings and transformer architectures to predict subsequent tokens, ChatGPT 01 Pro integrates real-time information retrieval, agentic reasoning, structured multi-step (chain-of-thought) reasoning, and a framework for external tool integration. These enhancements enable it to dynamically query updated sources (e.g., web searches or databases), invoke specialized computational resources (such as calculators or code execution environments), and methodically validate its outputs through retrieval-augmented generation, significantly reducing hallucinations and factual errors inherent in simpler models.

To systematically mitigate biases—far beyond standard training techniques—ChatGPT 01 Pro employs targeted methods such as Reinforcement Learning from Human Feedback (RLHF), careful dataset curation and balancing, explicit adversarial debiasing algorithms, model editing at the neuron or circuit level, and ongoing external fact-checking. In contrast to the implicit biases of human reasoning (such as confirmation bias or cognitive blind spots), these rigorous, engineered methods provide continual oversight and adjustment. Consequently, ChatGPT 01 Pro’s approach to managing biases is notably more deliberate, transparent, and self-correcting than typical introspective methods available to individuals, offering an analytical, structured alternative to subjective human judgment.

Posted
1 hour ago, Analytics said:

 

Yes, we all have our biases. The question is whether we embrace them because they provide the answers we prefer, or whether we actively seek to eliminate them because we prioritize knowing the truth—even when it causes tension with ourselves, our families, and our communities. Here is what ChatGPT 01 Pro does to systematically minimize its biases; how does that compare to your personal efforts to minimize yours?

ChatGPT 01 Pro, especially with “Deep Research” enabled, transcends being a mere large language model (LLM). While basic LLMs rely exclusively on pre-trained embeddings and transformer architectures to predict subsequent tokens, ChatGPT 01 Pro integrates real-time information retrieval, agentic reasoning, structured multi-step (chain-of-thought) reasoning, and a framework for external tool integration. These enhancements enable it to dynamically query updated sources (e.g., web searches or databases), invoke specialized computational resources (such as calculators or code execution environments), and methodically validate its outputs through retrieval-augmented generation, significantly reducing hallucinations and factual errors inherent in simpler models.

To systematically mitigate biases—far beyond standard training techniques—ChatGPT 01 Pro employs targeted methods such as Reinforcement Learning from Human Feedback (RLHF), careful dataset curation and balancing, explicit adversarial debiasing algorithms, model editing at the neuron or circuit level, and ongoing external fact-checking. In contrast to the implicit biases of human reasoning (such as confirmation bias or cognitive blind spots), these rigorous, engineered methods provide continual oversight and adjustment. Consequently, ChatGPT 01 Pro’s approach to managing biases is notably more deliberate, transparent, and self-correcting than typical introspective methods available to individuals, offering an analytical, structured alternative to subjective human judgment.

I know how LLMs work (I've coded most of the parts of what makes an LLM).  I know how debiasing works for LLMs (I've coded up adversarial algorithms).  As I said previously, they've done an amazing job.  They are still finding new and better ways to improve debiasing.  But if you are trying to convince me that I should trust the conclusions of an LLM over my own, you will not succeed.  I don't trust LLMs, because I know how they work.  LLMs are great as a research tool, great for bouncing ideas off of it, great for finding new ideas, but they are absolutely not great for coming up with conclusions.  Especially in areas that are not widely known.

Also, since the original chatgpt topic was about the Book of Mormon, I enjoy reading critical points of view to it.  I don't care about podcasts or videos so it is getting harder to find written critical points but I still search for them.  I've even used LLMs to find critical points.  I've always been open to reading critical points of view.  The first time that I can remember doing it was when I was in High School (it was about the peepstone).  My mother was worried about me.  I, instead, found it fascinating.

Posted
6 hours ago, Analytics said:

When talking about something so general, I don’t know what being “unbiased” would even mean. Whether biological or natural, the best any intelligence can do is make a good faith effort to look at all of the data, weight it appropriately, and decide how best to synthesize it to make decisions. From my experience, ChatGPT4.5 is less biased than any biological intelligence I’ve ever interacted with. 

Having said that, what do you think of its response? Do you think there is new or overlooked scholarship that it didn’t consider?

It is biased to be agreeable to whomever it is talking to.

Posted
On 3/17/2025 at 4:36 PM, Nofear said:

Ether's Elephant isn't exactly  a channel I would describe as TBM (often used as a pejorative by critics, but I consider myself a true believer) but it has fun points. This episode is kind of funny.....................

AI can be bullied into saying almost anything.  I had an AI lie to me.  When I told the AI that, it apologized.

Posted
12 hours ago, Analytics said:

To systematically mitigate biases—far beyond standard training techniques—ChatGPT 01 Pro employs targeted methods such as Reinforcement Learning from Human Feedback (RLHF), careful dataset curation and balancing, explicit adversarial debiasing algorithms, model editing at the neuron or circuit level, and ongoing external fact-checking. In contrast to the implicit biases of human reasoning (such as confirmation bias or cognitive blind spots), these rigorous, engineered methods provide continual oversight and adjustment. Consequently, ChatGPT 01 Pro’s approach to managing biases is notably more deliberate, transparent, and self-correcting than typical introspective methods available to individuals, offering an analytical, structured alternative to subjective human judgment.

The only challenge that generative AI faces in this particular arena though is that it doesn't do original research. When we get to issues like Mormonism, bias in the training material is a source of bias that a ChatGPT cannot easily eliminate. Anachronisms as discussed by ChatGPT above is an example of the problem because of the fact that there are embedded in virtually all LDS apologetic material and all critical material a shared set of assumptions that are (I have argued) incorrect. The Book of Mormon is a 19th century text - either as an original work or as a translation - and so we have to differentiate between anachronisms that can be attributed to the translation layer (assuming that there is one) and anachronisms that can't be when we are trying to compare the two theories. ChatGPT doesn't flag this sort of thing because in the literature it is trained on, this question is generally ignored. This is not to say that anachronisms don't exist - merely that our discussions of anachronisms are generally engaging a range of assumptions that are never really spelled out (except in the unrelated specialist literature of translation itself - where I have only found one source that discusses the Book of Mormon in terms of translation theory), and so largely unavailable to ChatGPT when engaging this sort of question.

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...