Transcription
I pulled five specific claims from Dan's video, Why You Can't Trust the Bible. Five times he invoked the overwhelming majority clause, and I ran them through AI. All right, let's see it. Let's take a look. Well, you're thinking of the New Testament, which was written fully within 70 years of Jesus's death and resurrection. Uh, this is another dogma that's not supported by the data that presupposes every word of the text is true. The overwhelming majority of scholars agree that there are parts of the New Testament that were not written until the 2nd century CE, which is over 70 years after Jesus's death. Dan's second claim: the overwhelming majority of scholars agree that parts of the New Testament were written well into the second century.
So, I have two main concerns with this video. And the second one has to do with the laughable notion that AI is somehow a competent arbiter of the scholarly consensus. But before we get to that, we need to deal with the egregious misrepresentation of my claim. Because my claim was the following: "The overwhelming majority of scholars agree that there are parts of the New Testament that were not written until the second century CE." That's not the claim that this content creator repeated after the completion of my video. The claim that this content creator repeated after the completion of my video was the same claim that was fed into the AI chatbot. In fact, in this image, you can see the text of the claim this content creator attributes to me: "The overwhelming majority of scholars agree parts of the NT were written well into the 2n century," which is something that I absolutely never said anywhere in that video. This is a flagrant misrepresentation of what my claim was. Now, this might strike you as kind of small potatoes, as kind of splitting hairs, but this content creator later on in this very video is actually going to distinguish the claim falsely attributed to me from a more reasonable claim, which is actually the claim that I made. So, this content creator is flagrantly misrepresenting my claim from the start. That is what is fed into the AI chatbot, which means that the results of that are not going to be representative of the scholarly position relative to my actual claim. So, right off the bat, the entire video is undermined, and this content creator's competence to engage specific claims is thrown into significant doubt. But let's get to the next part.
So I asked AI, "Is that the majority position?" Here's what it came back with. Even among critical scholars, Dan's own camp, the majority position dates the entire New Testament cannon is completed by around 100 AD. That's not a conservative talking point. That's the critical mainstream.
So, according to this content creator, Claude says that it is the critical mainstream position that all 27 books of the New Testament were completed by the year 100 CE. And that is definitely not true. Having read as much of the scholarship as I can get my hands on regarding the dating of the New Testament, it is definitely a majority position among critical scholars that at least one text of the New Testament was likely written in the 2 century CE. Even among conservative scholars, there is recognition that if the pastoral epistles and Second Peter are pseudopagraphic, then the window of dating extends into the 2n century CE, and in some cases, even well into the 2n century CE. So, for instance, Jonathan Bernier, or Bernier if you're nasty, recently published *Rethinking the Dates of the New Testament, the Evidence for Early Composition*, and will have out shortly a handbook of New Testament chronology. And he points out, "Look, if Second Peter and the pastoral epistles are pseudapographic, if they were not written by Paul and Peter, then that window of chronology extends from around 60 CE to 125 for Second Peter, to 150 for 1 and 2 Timothy, and to 175 CE for Titus." A lot of scholars recognize it is likely that at least one text was written in the 2 century CE, if not more. The notion that this isn't even a representative or a majority view among critical scholars is just laughable.
Now again, AI is not a competent assessor of scholarly consensus because they don't have access to the overwhelming majority of the scholarship. What they have access to is what people on the internet are saying about the scholarship or whatever scholarship is publicly available or whatever scholarship has been published by publishers who decided to make their scholarship accessible to LLMs. And this is a big debate within scholarship. Well, it's not really so much of a debate. Scholars are pretty unilaterally against it. It is publishers who want to make money who are not going to be paying those scholars for their intellectual property who are turning it over to LLMs. But I decided to test out the AI chat bots that I'm familiar with on these claims and mainly my actual claim and then where they're getting this information from. And I asked Claude, I asked Chat GPT, I asked Grock, and I got three very different answers that run a spectrum. But when I asked if they were getting their answers from a review of the actual scholarship or from a review of what is publicly available on the internet that has been said about the actual scholarship, every last one of them had the exact same answer. But let's take a look at what I did.
So the question I asked was this: "Would it be accurate to state that the overwhelming majority of biblical scholars date the composition of at least some texts of the New Testament into the 2 century CE?" Claude's response was, "No. That wouldn't really be accurate." And it's actually kind of the opposite of the scholarly consensus. So, this answer is in alignment with the answer that this content creator got, even though they asked a completely different question. Claude goes on to say, "The overwhelming majority of critical biblical scholars date the core New Testament text to the first century CE, roughly between the 50s and the '9s/ early 100s CE." Now, early 100s CE would be 2 century CE. But, uh, we go down to, "What about second century dates?" A minority of scholars do argue that certain texts, most commonly Second Peter, the pastoral epistles, 1 and 2 Timothy, Titus, or occasionally the Gospel of John, were composed in the very early 2nd century. But this is a minority position for most of those texts. And even where it's held, it's usually talking about the very early 100s CE, not a dramatically late dating. So this is not true.
Uh, now, there's no button we can push to get an accurate assessment of where all scholars stand on any given research question. The best you can do is read as much of the scholarship as you can find and then figure out where the majority of scholars stand. Now I'm working on a large-scale state of the field biblical studies survey. I have been for some time now. It's a little more complicated than I thought it was going to be. I get asked about it all the time, but, uh, it's moving forward, and it's moving on to the next stage, uh, in the process of getting everything taken care of. But this is one of the questions that's on that survey along with a number of others. So I'm going to have very robust data on this, better data than anything AI has access to, better data than have ever been provided before. So that will be forthcoming, and so I'll be able to state with a lot more certainty the exact numbers before too long. But for now, the best we can do is read all the scholarship and count up the positions that are held. AI cannot do that. AI can only see what people on the internet have said about it. And they don't really accurately distinguish between Reddit and blog posts and discussion boards and actual scholarship that has been published.
Now, I asked Chat GPT the same question. Here's how they responded: "s with one important qualification. It would be accurate to say something like the overwhelming majority of critical biblical scholars state the composition of at least some New Testament writings to the 2 century CE. If you broaden the pool to include conservative evangelical and confessional scholars, the claim becomes a little less categorical. But it is still likely accurate to say that most biblical scholars overall, except a second century date for at least one New Testament text. I would hesitate only over the phrase overwhelming majority unless you explicitly mean the critical scholarly mainstream." So, Chat GPT says, "Yeah, the characterization is a little strong, but mainly, yeah, that's accurate."
Then I asked Grock the exact same question. Grock's response: "Yes, it would be accurate. Mainstream biblical scholarship across critical, moderate, and many conservative views dates the composition of the New Testament books primarily to the mid to late 1st century CE, roughly 50 to 100 CE, but with important qualifications that support the statement. You got to go down to the second bullet point. Disputed later text: Several books have majority or strong scholarly support for dates in the early 2nd century or ranges crossing 100 CE. These include Second Peter. The large majority of critical scholars view it as pseudographical and date it to the first half of the 2n century, often between 100 to 150 or 160 CE, making it one of the latest books, and then the pastoral epistles." So the three different chat bots I asked had three different answers. One said, "Haha, no." One said, "Yeah, but we got to soften the rhetoric a little bit." One said, "Yes, you are correct."
Now, I asked all of those chat bots this question: "Is your answer to this question based on a survey of a wide range of peer-reviewed academic publications, or is it more directly based on publicly available comments regarding the trends of peer-reviewed academic publications?" Claude said, "This is a really fair and important question to ask, and you deserve an honest answer. My response is based on the second category. I wasn't trained by systematically reading and analyzing a comprehensive corpus of peer-reviewed biblical scholarship journals like New Testament studies, Journal of Biblical Literature, etc. My knowledge reflects general representations of scholarly consensus as they appear in textbooks, encyclopedias, popular academic writing, and publicly available commentary about the field." This is Chat GPT's answer: "The latter. My previous answer was not based on my having conducted a systematic survey of a representative sample of peer-reviewed literature on New Testament dating. Rather, it was based on my internal knowledge of the scholarly landscape, which itself reflects patterns learned from a mixture of sources, including textbooks, reference works, peer-reviewed scholarship, publishers descriptions, public discussions by scholars, and other publicly available material used in model development." Grock answered the same way: "It is more directly based on publicly available summaries, overviews, and discussions of trends in peer-reviewed academic publications and scholarly consensus."
So AI does not have direct access to the overwhelming majority of peer-reviewed published scholarship. If it is in the public domain, it has access. If they have signed a contract with a publisher to turn over their materials for an LLM, it has access. Otherwise, it's looking at blogs, it's looking at podcast interviews, it's looking at Reddit, it's looking at discussion boards, it's looking at social media to figure out what people are saying, and it's not the best at discerning expert opinion from non-expert opinion. It even flagrantly misrepresents scholarship it does have access to. I have frequently tested out AI by saying, "Name the publications from the last 30 years that endorse this particular position on a given research question," and it'll rattle off a bunch of publications. Some of them are just hallucinated. They don't exist. Sometimes they will attribute a position to a scholar who argues for the exact opposite position in the very publication they point you to. They make all kinds of mistakes and they don't have access to most of the scholarship. Again, at the moment, the only reliable way to discern what scholars think is to read their publications, which is very time consuming, which is very costly, and most people do not have the time, the access, or even the interest to do so. I am trying to provide a tool for exactly that, uh, in the future, and hopefully by the beginning of next year, we'll have results that we can share.
So asking AI if my claim was right is problematic for a number of reasons, and that's setting aside the fact that the claim was flagrantly misrepresented. Having said that, I am happy to acknowledge that my characterization of the field that agreed with that second century dating as the overwhelming majority could have been overstating the case a bit. I would say a majority or a strong majority. In fact, here's another video from somewhere close to the same time period where I discuss this issue. And here's how I put it in that video: "There are a minority of scholars that greatly exceeds 1% that would argue that the dating window of texts like the Gospel of Luke, or the Gospel of John, or Second and Third John, or even Revelation, extends into the first or even the second quarter of the 2 century CE. And you've got other texts like the pastoral epistles where that percentage is probably more like 25 to 50% of scholars would say the dating window extends into the first or the second quarter of the 2nd century CE. And when it comes to texts like Second Peter, I think the majority of scholars agree that the text was written in the 2n century CE. Even conservative scholars will say the dating window goes into the first quarter. But I think the majority of scholars would say it was probably written between like 100 and 150 CE." So I would say that's a more judicious approach. And I'm happy to accept and acknowledge the criticism that I got out a little over my rhetorical skis with that claim. And I'm not saying this to just dismiss it and move on. This is something that has been pointed out on occasion to me. And I am aware that sometimes that does happen. So, I'm happy to accept that criticism and thank this content creator for bringing my attention to that even if the overwhelming majority of this video is based on misinformation and a ridiculous falsehood, namely that AI is a competent arbiter of the scholarly consensus.
So, my two main concerns here, and we're going to get on to some more smaller concerns, but my two main concerns here are: one, my claim is flagrantly misrepresented; and two, AI is not a competent arbiter of what the majority of scholars think.
Now, the books of the canon that get pushed latest are the pastoral epistles, First and Second Timothy, and Titus. Critical scholars do debate those, but even the most aggressive critical dating for those books lands them at early 2nd century at most, not well into the second century. So here we have a distinction being drawn between the most aggressive critical positions and my claim that they date well into the 2n century CE, which again was not my claim. In fact, my claim was that it's in the early 2nd century CE. But also, this is a misrepresentation of the critical scholarship and likely based on this content creator's reliance on AI to adequately represent the scholarship, because even conservative scholars such as Bernier and others who provide a window of potential dating for things like the pastoral epistles and Second Peter. Even Bernier says that the pastoral epistles, for instance, if they are pseudopographic, the window goes into the second half of the 2 century CE, to 175 CE for one text. So it is not accurate to say even the most aggressive critical dating only places it in the early 2nd century CE. It seems to me, and now this is based on my review of the scholarship, that the majority of scholars would say Second Peter at least was probably composed in the 2 century CE. And my review of the scholarship suggests that you can whittle down the majority of the windows to somewhere around 120 to 150 CE. It seems to me that's where the most scholars are suggesting it was likely written. Now, I could be very wrong about that, but if it's not where they suggest it was written, it's at least, uh, a significant overlap of all of the different chronological windows. But, uh, the claim that, no, even the most aggressive dating only goes into the early 2nd century CE, is inaccurate according to my understanding of the scholarship, which, by the way, is not just relying on what an AI chatbot had to say. And conservative scholars like Carson, Guthrie, Moo have produced serious peer-reviewed academic work defending first century dates for every single book in the canon. These aren't popular level apologists. These are credentialed scholars engaging the same evidence, and those are minority positions that are heavily criticized by other scholars.
So, "well into the second century" overstates even what the critical mainstream actually claims. So, in addition to the fact that, again, this is not my claim, this is just punting to one AI chatbot who got it wrong. Dan pushed past his own camp's majority position. The verdict: his stance doesn't hold up. The verdict? Don't use AI to try to adjudicate the scholarly consensus. And also, uh, here's an image of what this content creator ended with: "Don't yell at me, yell at Claude." And Claude is spelled incorrectly, but you flagrantly misrepresented my position. So that's on you. You also relied on an AI chatbot to do something that an AI chatbot is not designed to do and can't competently do. That's also on you. And as we've seen, Claude got stuff significantly wrong, but the different chat bots run a gamut of responses to this question. And so if you want to understand where the scholarship stands, go read the scholarship, which this content creator has not done, and also will not do. And the fit for this video has been Lobo.