Syntagma Theologiae Christianae - AI Translation

Status
Not open for further replies.

dsanch1120

Puritan Board Freshman
Hello everyone,
I was inspired yesterday to try and automate a translation of Polanus' "Syntagma Theologiae Christianae" using some AI and ML tools that I have at my disposal (I'm a software developer). I've thus far attempted translations of the epistle to the reader and the synopsis of book 1. The process still needs some fine-tuning and I've had to go through and modify a few things throughout, however, the translations generated seem to be decent.
I've posted the translations here
- Epistle to the Reader
- Book 1 Synopsis
I would love to hear anyone's thoughts, feedback, and/or criticism on this effort. I would love to continue the translation efforts for as long as I'm able to do so as it doesn't take a huge amount of effort or cost per page. The difficulty is that there's over 4000 pages, so it would take quite a long time. In any case, I've enjoyed what I've read of AI Polanus so far.
 
Some of our PB Latinists should weigh in. @Charles Johnson I know that from work I'm doing, except for short phrases (and sometimes not that), automated is not adequate and by that I mean professionally to formally publish. Informally, studying, to get an idea, maybe; but I'm not a Latinist. Even Latinists can get things wrong; so surely AI is not bullet proof; there's an example that came up on the board about a passage in Rutherford from a formal publication of a portion of one of his works from a few years ago.
 
To be honest with you, I wouldnt waste my time with it at the moment. AI is not good enough yet to translate this much Latin. Maybe small phrases, but not a 4,000 page book. Nobody would really buy it, as they cannot quote it, nor be sure it is Polanus' genuine position(s.) Your heart is in the right place, but this is a pretty serious work to leave completely in the hands of AI in its current state, especially not knowing Latin yourself. It would be different if you knew Latin, and were using it as an aid, but you are going in it blind.

But if you feel strong about it, do it. Do not let anybody dissuade you. If I were you though, I would give AI another 5 years at least and revisit this plan. It will be many times more proficient by then. But again, do not let people dissuade you if you feel God is calling you to do this. He often works in mysterious ways. I would just hate to see you do all this effort, and in 5-10 years, AI advance so much, that you will be able to simply upload the entire PDF and it spit it back out fully transcribed and translated. AI is eventually going to revolutionize the translation game as well. Just got to give it some time.

I was initially more encouraging in my former response, but after reading your "about" on your website, this looks like it is going to be kind of a side project; and one you are not fervent about at that. With this in mind that is why I say dont even bother. Because that page or two, here or there, probably spanning years; by the time you spend those years making progress, AI will be to a point it can do what you did in those years, instantaneously at best, or in a few days at worst. Enjoy yourself, keep the project in mind, and revisit it at another time. AI is probably going to get to the point where anybody can do projects like these with relative ease, and as such, there will no longer be much of a market for them to be that profitable.

But again, do not let anybody dissuade you if you feel strongly about it. I know I didnt when we were brainstorming about Project Puritas, and it has allowed almost 100 Puritan classics to be digitally rePub'd and freely dispersed in the last year. Sometimes the best things come from dreams that nobody believes in; and its even better when you find a couple people that do. If you do decide to go through with this, let me know, I will help in any way I am able. But it will be like you, when time allows.
 
Last edited:
This is a very bad idea. The AI didn't even get the first sentence correct (evidently it couldn't parse the Greek word "ομωνυμως").

I'm sure your heart is in the right place brother, but when AI Polanus gets to matters like the doctrine of God, or justification, or heaven or hell, you're going to have no way of knowing that it's not spitting out heresies.
 
This is a very bad idea. The AI didn't even get the first sentence correct (evidently it couldn't parse the Greek word "ομωνυμως").

I'm sure your heart is in the right place brother, but when AI Polanus gets to matters like the doctrine of God, or justification, or heaven or hell, you're going to have no way of knowing that it's not spitting out heresies.
Thank you for your response, I appreciate the honesty. I was hoping to hear whether it was accurate or not, and now that I know, I’ll take down anything public and likely won’t continue to use this personally either based on the feedback I’ve gotten. You all make good points, that I should have considered more, so thank you!
On the bright side, I can probably utilize some of the experience from building the technological backend in my job, so not a completely wasted effort.:)
 
Your response was thoughtful, but I'd quibble with the timeline. I'd try again every 6 months, a year at most if one was serious.
I’ll be keeping an eye out. I think a large issue with it was the inability to transcribe Greek (and presumably Hebrew), which I imagine won’t be as much of a problem as time goes on.

mistakes. If you wanted to still contribute something that could benefit the church, there is still plenty left to do. Knox's Updated, or Musculus' both would be good even if you wanted to publish professionally.
Yeah that definitely would be good to look into. Are you thinking of latin to English, or more so older English to modern English with notes added?
 
I’ll be keeping an eye out. I think a large issue with it was the inability to transcribe Greek (and presumably Hebrew), which I imagine won’t be as much of a problem as time goes on.


Yeah that definitely would be good to look into. Are you thinking of latin to English, or more so older English to modern English with notes added?
Sorry, I deleted my post, didnt know if you were wanting to keep it focused on Polanus. No, both Musculus and Knoxs works are older English "types." Knox's is archaic Scottish, while Musculus is Old English with the facsimile in Blackletter (which makes it hard to read for most.) The one complaint I see of Knox's works across the board is Banner reprinting it in Gaelic instead of doing a complete revision to modern English. What most people are saying is it is good for reference, but very poor for devotional or casual reading. Chat did an excellent job of updating the Old Scottish to contemporary English. What I was mentioning about having the Gaelic on top, and the translation on bottom, is you will have a comparative output that will make it easy to update. If it is sentence by sentence, or paragraph by paragraph, yourself, or a crowd effort, can easily fine tune it to make the minute corrections that will need to be made on its way to publication. All you have to do is ask Chat to process it this way, i.e. Original on top, translation on bottom, sentence by sentence, and it will spit it out as such. People would be more willing to trust an English work from an English speaking person. They would trust it even more if the original were included in some way, and you gave yourself time to receive some peer feedback for corrections. Just an idea. If you have already built a model, there may still be much use for it. Some of this stuff people just dont want to deal with. Logos didnt even want to update it, they made their module out of David Laings 19th century version. But Chat now processes a translated page in seconds, while it may take a few days from a professional publisher. I would attempt it, but I dont have the free time to dedicate the month or two to do all 6 volumes.

Seriously though, if you changed that website to updating Knox, and used each page to post segments with the original on top, and the translation on bottom, while using the comments section as a way to receive peer feedback, it may just be pretty active. And you can still publish each volume in your mode of choice. The only danger though, is since the output will have come from AI, there may be a question of your ability to copyright it as a creative work, which may damper your chances of securing protection against unintended distribution. I dont know much about copyright law because the works I do are in the Public Domain, and offered freely, so we hope people share as much as possible, but just dont copy and try to sell. So there is a chance any major effort you do with AI, simply cant be copyrighted. While there is a blessing in what it can do, many people are going to shocked to find out their claims getting shut down thinking prolific use of it constitutes a creative work, or is going to be ultimately protectively profitable. Many Christians will not push the issue, as we look to it as love for neighbor regardless what the law allows; but some may. Then the question becomes are they actually sinning if working within the realm of national laws? The same goes for most all republications of Public Domain works (anything pre-1930 in the US.) Somewhere along the line, Copyright Law decided that updating, or putting Public Domain works in a different medium does not constitute a creative work; and one has to prove a major revision to do so; not just in spelling, formatting, or mode, but in actual content. We should all know that going in.

A translation would count as a creative work. An AI translation would not.

copyright law.jpg
 
Last edited:
All you have to do is ask Chat to process it this way, i.e. Original on top, translation on bottom, sentence by sentence, and it will spit it out as such
Oh yeah that’s 100% doable. I think I’d just have to change 2-3 lines of code, I’ll think about that.
 
Oh yeah that’s 100% doable. I think I’d just have to change 2-3 lines of code, I’ll think about that.
Hey, I just wanted to give you the heads up, that I talked to my boss, and he has given me a month to see what I can do with the 6 volumes. Just kind of feel that this is something that will benefit the Church Universal, and dont want it to be restricted necessarily, to a need to be purchased access. John Knox's Works have been kind of stuck in their Early Scottish/English dialect for a long time, and this is an opportunity to try and free them. With that being said, if you still wanted to do a "critical updated edition" there is room for that. The way I plan on doing his Works will be more like "A Reader" or "An Anthology" bringing them more into a devotional perimeter than scholastic or reference. Seeing that the standard Laing copy is freely accessible online, and that I think Banner's even follow the same pagination (though I could be wrong on that) we will include the page numbers for each section of text correlating with that edition by a bolded and super-scripted indication. We are doing this also, because we are focusing solely on "the Works of John Knox" and not the footnotes, which sometimes can take up almost an entire page. Including the page numbers to the easily accessible original version, will make referencing those footnotes, or comparing the update per section, pretty easy if any questions arise. This will also shrink his actual works considerably, which may make low-cost PB's a reality too. I will still be going over the text word by word to check it out; and the goal is not to update to modern style, just modern spelling.

There is no competition brother, in honesty, if you wanted to, you can do the exact same thing and charge for it, the text is Public Domain. Was just letting you know it is now being attempted, with the goal of distributing it at no cost.
 
Last edited:
There is no competition brother, in honesty, if you wanted to, you can do the exact same thing and charge for it, the text is Public Domain. Was just letting you know it is now being attempted, with the goal of distributing it at no cost
Good to know! I don’t care much about charging for it if it would benefit the church. If it would be helpful, I can DM you more specifics about the process I used.
 
Knox's is archaic Scottish, while Musculus is Old English with the facsimile in Blackletter (which makes it hard to read for most.) The one complaint I see of Knox's works across the board is Banner reprinting it in Gaelic instead of doing a complete revision to modern English.
Just to clarify, Gaelic is a completely different language, not archaic Scots. It's extremely unlikely Knox would have spoken Gaelic, which would likely have been restricted to the Highlands and Islands of Scotland during that time. Dunkeld, in Central Scotland just north of Perth, is about as far south as Gaelic would have been widely spoken.
 
Just to clarify, Gaelic is a completely different language, not archaic Scots. It's extremely unlikely Knox would have spoken Gaelic, which would likely have been restricted to the Highlands and Islands of Scotland during that time. Dunkeld, in Central Scotland just north of Perth, is about as far south as Gaelic would have been widely spoken.
Thank you for giving the heads up. Chat responded with that when I asked it to update the text from Gaelic to English. Something along the lines "actually, the text is not Gaelic, it is a form of Early English with Scottish influence." Then kind of gave me the history of it. Now the only command I have to ask for each page is "can you update this page to English without changing the phrasing." And it updates it to modern English without trying to rephrase everything to make it read like a modern book. If it did that, it would be hard for me to follow along when I proofread it, as the text wouldnt align.

Good to know! I don’t care much about charging for it if it would benefit the church. If it would be helpful, I can DM you more specifics about the process I used.
Thank you for the offer. I am not too tech savvy. Found out today that Chat 4o will (supposedly) let you do 80 requests every 3 hours, so this should work fine getting the 3,600 pages or so of all the volumes into word. Should take me, Lord Willing, only about 15 days or so. Like I was mentioning, the size reduces dramatically when you dont include the footnotes, for instance right now I have 16 pages in from the source, but only 6 in Word at 11pt font.

You may try to explain it to me, but I would probably be lost, lol. Thank you again for the offer though.
 
Last edited:
Status
Not open for further replies.
Back
Top