Latin Reformed Vocabulary Builder? Good Idea or No?

[davejonescue]

So today I was thinking something, as I am trying to come up with ideas possibly of getting closer to reading fluency in as little as time as possible. Of what I understand it, while grammar rules are important, so is building vocabulary. This is what got me thinking, when attempting to transcribe the text for translation, the hard part is cleaning the facsimiles and looking for mess-ups. But, what if instead of trying to produce translations, the goal is to transcribe simply to get a list of unique words and word frequencies? So, say I transcribe what are known as the "pivotal" Reformed Scholastics, (if I can find a list), generate a word list; and then, turn those words into phonetic renderings as to be able to create audio flash cards to listen to, on the words (starting from the most used, to the least) from those authors?

There are a few reasons why this is possible:

1. Claude is at 99% accuracy.
2. Most of the matching, phonetic conversion, frequency counts can be automated.
3. This will potentially get folks prepared to read "the majors" by knowing the vocab going in.

Someone on here said that there are only so many 1000's of words one has to have down to get going really good. I can't find that list. I was able to get all the words and definitions from something called Latdic, (which has a public domain notice) but that consists of about 40,000 words, which seems impossible to memorize (and many are probably not needed). I don't think it would be hard to compare the transcripts list to that list, see what there are already definitions for (as to create either digital/audio flashcards) and find the definitions to the words that are not there.

Just seems like a way to take the present tech capabilities and give a little push in the right direction to being able to read these books fluently when/if someone ever attempts to read them.

What do you guys think?

 

[jhcampbell]

So today I was thinking something, as I am trying to come up with ideas possibly of getting closer to reading fluency in as little as time as possible. Of what I understand it, while grammar rules are important, so is building vocabulary. This is what got me thinking, when attempting to transcribe the text for translation, the hard part is cleaning the facsimiles and looking for mess-ups. But, what if instead of trying to produce translations, the goal is to transcribe simply to get a list of unique words and word frequencies? So, say I transcribe what are known as the "pivotal" Reformed Scholastics, (if I can find a list), generate a word list; and then, turn those words into phonetic renderings as to be able to create audio flash cards to listen to, on the words (starting from the most used, to the least) from those authors?

There are a few reasons why this is possible:

1. Claude is at 99% accuracy.
2. Most of the matching, phonetic conversion, frequency counts can be automated.
3. This will potentially get folks prepared to read "the majors" by knowing the vocab going in.

Someone on here said that there are only so many 1000's of words one has to have down to get going really good. I can't find that list. I was able to get all the words and definitions from something called Latdic, (which has a public domain notice) but that consists of about 40,000 words, which seems impossible to memorize (and many are probably not needed). I don't think it would be hard to compare the transcripts list to that list, see what there are already definitions for (as to create either digital/audio flashcards) and find the definitions to the words that are not there.

Just seems like a way to take the present tech capabilities and give a little push in the right direction to being able to read these books fluently when/if someone ever attempts to read them.

What do you guys think?

While the idea itself isn't bad, it is not as if you can use osmosis to put the words in your brain. I think it would be better to read naturally and then learn the meanings of those words as they come along, in the context of a real book, instead of trying to slam them all in your head with flash cards

 

[davejonescue]

While the idea itself isn't bad, it is not as if you can use osmosis to put the words in your brain. I think it would be better to read naturally and then learn the meanings of those words as they come along, in the context of a real book, instead of trying to slam them all in your head with flash cards

I am really thinking of using them for passive/ambient learning. i.e. making audio flash cards from them to use when I go on walks, work, work-out, etc. I have checked into TTS, and while I cant find any specific one programmed for Latin, it does appear if you convert the Latin into phonetic spelling, pretty much any TTS should be able to render it. This is why I would first want to compare the frequency list to the LatDict list I already have as they come with definitions.

Another thing is, once you have a list of the collective words, and their frequencies, you should be able to make books with just those words, so that, say reading this or that story, prepares you to read so many other authors.

Lastly, the above two are just the tip of the iceberg of ways to get to learn the words. One of the good things with AI, is once you have the list, you could learn it in tons of ways, i.e. drills, games, etc. In reality, you could probably vibe-code a few pretty cool games to learn it in your free time.

Again, just an idea.

 

[RamistThomist]

Just memorize hundreds of Latin flash cards. It's easier than it sounds. And it will pay off for actually reading Latin.

 

[NaphtaliPress]

You all may be ahead of me. I don't know Latin but you need grammar as well as vocabulary. I queried AI and it remarks, " If you only memorize simple definitions (like "puer = boy"), you’ll get stuck as soon as you see puerum, pueri, or puero in a sentence. Think of vocabulary as the building blocks and grammar as the blueprints. You can’t build a house without both.

How to use flashcards effectively for Latin:
Instead of just one word per card, seasoned learners use specific layouts to bridge the gap between "vocab" and "grammar":

• Nouns: Don't just write puer. Write the nominative, the genitive, and the gender (e.g., puer, pueri, m.).
  • Why? The genitive singular (pueri) tells you exactly which "rule set" (declension) the word follows.
• Verbs: Write all four principal parts (e.g., voco, vocare, vocavi, vocatum).
  • Why? These give you the "stems" needed to build every possible tense, from "I call" to "I had been called".
• Adjectives: List the masculine, feminine, and neuter endings (e.g., magnus, -a, -um).
  
  A better strategy for beginners:
• Learn in context: Use flashcards that have a full sentence on the front with one word missing (called "cloze deletion"). This helps you learn the meaning and the grammar at the same time.
• Use the "Core" first: Focus on the Dickinson College Commentaries (DCC) core list of the 1,000 most common words, which make up about 70% of most Latin texts.
• Combine tools: Apps like Anki or Clozemaster are popular because they use "Spaced Repetition" to show you words right before you're about to forget them." For what it's worth.

 

[davejonescue]

You all may be ahead of me. I don't know Latin but you need grammar as well as vocabulary. I queried AI and it remarks, " If you only memorize simple definitions (like "puer = boy"), you’ll get stuck as soon as you see puerum, pueri, or puero in a sentence. Think of vocabulary as the building blocks and grammar as the blueprints. You can’t build a house without both.

How to use flashcards effectively for Latin:
Instead of just one word per card, seasoned learners use specific layouts to bridge the gap between "vocab" and "grammar":

• Nouns: Don't just write puer. Write the nominative, the genitive, and the gender (e.g., puer, pueri, m.).
  • Why? The genitive singular (pueri) tells you exactly which "rule set" (declension) the word follows.
• Verbs: Write all four principal parts (e.g., voco, vocare, vocavi, vocatum).
  • Why? These give you the "stems" needed to build every possible tense, from "I call" to "I had been called".
• Adjectives: List the masculine, feminine, and neuter endings (e.g., magnus, -a, -um).
  
  A better strategy for beginners:
• Learn in context: Use flashcards that have a full sentence on the front with one word missing (called "cloze deletion"). This helps you learn the meaning and the grammar at the same time.
• Use the "Core" first: Focus on the Dickinson College Commentaries (DCC) core list of the 1,000 most common words, which make up about 70% of most Latin texts.
• Combine tools: Apps like Anki or Clozemaster are popular because they use "Spaced Repetition" to show you words right before you're about to forget them." For what it's worth.

Yes, it is said you do need grammar with the vocabulary. But it is hard for me to believe word list percentage estimates that were produced before Latin could be OCR'ed to a good deal of accuracy. Meaning, how many Reformed works have been digitized and compared? Now, I don't really have to guess. I can auto-transcribe Mastricht, Voetius, Vitringa, DeMoor, Heidegger, Polanus, etc. or those with large "Reformed Systems", and focus on the vocabulary within them. Not only that, but once the list is had, one can use AI to produce all sorts ways to learn it; from guides, cards, audio, short-movies, etc. My thing is, a lot of people want to tout something pubished 70 years ago as the best method to presently learn Latin. Its hard for me to see how that could be the case? You said you think it better to deal with the 16th and 17th centruy vocabulary, as it is written by them? This is doing that.

Of what I understand, yes, you are correct Grammar is important, but, so is vocabulary. Why would I want to focus on vocabulary that is not in the works I want to read?

 

[jhcampbell]

Of what I understand, yes, you are correct Grammar is important, but, so is vocabulary. Why would I want to focus on vocabulary that is not in the works I want to read?

For me, the aim of my goal in learning Latin (which is different from yours) is fluency in general, not just in the books I desire to read. That way I can read broadly, not just where I have been trained to read. What if a reformed book quotes something from somewhere else, and I want to read it, but I don’t have that vocabulary because I only focused on reformed works? That’s why my aim is fluency in general

 

[davejonescue]

For me, the aim of my goal in learning Latin (which is different from yours) is fluency in general, not just in the books I desire to read. That way I can read broadly, not just where I have been trained to read. What if a reformed book quotes something from somewhere else, and I want to read it, but I don’t have that vocabulary because I only focused on reformed works? That’s why my aim is fluency in general

I see that point. How long do you think it will take you to learn every Latin word? For me, I think that would take quite a while. If I know my goal is to read the Reformed systems, and by isolating the vocabulary, see what words they were using, then that will get me off to a good start. If I am only fluent in them, I will be content.

 

[jhcampbell]

I see that point. How long do you think it will take you to learn every Latin word? For me, I think that would take quite a while. If I know my goal is to read the Reformed systems, and by isolating the vocabulary, see what words they were using, then that will get me off to a good start. If I am only fluent in them, I will be content.

Fluency isn’t about every single word. That’s kind of a false setup. I’m fluent in English but I know for sure I don’t know EVERY WORD

 

[davejonescue]

Fluency isn’t about every single word. That’s kind of a false setup. I’m fluent in English but I know for sure I don’t know EVERY WORD

You most likely know every word that is in frequent usage in the level of reading you most frequent. This is the same idea with pin-pointing the words in the text I want to read. They become frequent to me before I engage the text, so that, L.W. when I engage the text, the reading is pleasurable and edifying, and not a burdensome task in itself. Its not worth the effort, with so many English books, if I have to start, stop, start, stop, start, stop, etc. To me, there is no glory in knowing the language. Its only what knowing the language allows one to read.

But to be perfectly honest with you, now that I am thinking about it. I should probably just wait five years if the Lord allows me to live that long, and something like all this would have already taken place. There are millions/billions of Christians. Certainly this isn't a novel idea. I don't know. We will see.

 

[RamistThomist]

Of what I understand, yes, you are correct Grammar is important, but, so is vocabulary. Why would I want to focus on vocabulary that is not in the works I want to read?

Because, as Chris mentioned, the declensions indicate whether a word is a direct object, subject, indirect object, genitive, etc. Vocabulary by itself doesn't tell you that. It just gives you the subject.

 

[NaphtaliPress]

See Jacob's comment about the difficulty about vocabulary cards said better than I did. Conceivably then, you create that list and then go back and pull the full sentences for each use in context to build cards from those Latin works? Okay. But it just seems to me that you want to skip training wheels (simpler sentences / classical start) and go to the deep end of the pool (to mix metaphors), and go to those Reformed theological works immediately. Voetius I know can have very difficult constructions. And Rutherford; Rutherford's Latin is no better constructed than his English (according to Dr. Noe) in his scholastic works. These are very difficult Latin works.

Yes, it is said you do need grammar with the vocabulary. But it is hard for me to believe word list percentage estimates that were produced before Latin could be OCR'ed to a good deal of accuracy. Meaning, how many Reformed works have been digitized and compared? Now, I don't really have to guess. I can auto-transcribe Mastricht, Voetius, Vitringa, DeMoor, Heidegger, Polanus, etc. or those with large "Reformed Systems", and focus on the vocabulary within them. Not only that, but once the list is had, one can use AI to produce all sorts ways to learn it; from guides, cards, audio, short-movies, etc. My thing is, a lot of people want to tout something pubished 70 years ago as the best method to presently learn Latin. Its hard for me to see how that could be the case? You said you think it better to deal with the 16th and 17th centruy vocabulary, as it is written by them? This is doing that.

Of what I understand, yes, you are correct Grammar is important, but, so is vocabulary. Why would I want to focus on vocabulary that is not in the works I want to read?

 

[davejonescue]

Because, as Chris mentioned, the declensions indicate whether a word is a direct object, subject, indirect object, genitive, etc. Vocabulary by itself doesn't tell you that. It just gives you the subject.

See Jacob's comment about the difficulty about vocabulary cards said better than I did. Conceivably then, you create that list and then go back and pull the full sentences for each use in context to build cards from those Latin works? Okay. But it just seems to me that you want to skip training wheels (simpler sentences / classical start) and go to the deep end of the pool (to mix metaphors), and go to those Reformed theological works immediately. Voetius I know can have very difficult constructions. And Rutherford; Rutherford's Latin is no better constructed than his English (according to Dr. Noe) in his scholastic works. These are very difficult Latin works.

Here is my thing. It is hard for me to believe that Lingua Latina, a book published 70 years ago, is still the best tool we have today, in 2026, for learning Latin. I know that's how some did it, but the reality is, AI does change a lot of things. Figuring out how to implement AI and the older methods to come up with the best current method is the hard part. I guess I will have to research it more.

 

[Venator Veritatis]

Here is my thing. It is hard for me to believe that Lingua Latina, a book published 70 years ago, is still the best tool we have today, in 2026, for learning Latin. I know that's how some did it, but the reality is, AI does change a lot of things. Figuring out how to implement AI and the older methods to come up with the best current method is the hard part. I guess I will have to research it more.

You're shooting in the dark vs a tried and true method to actually make progress. AI does not change the fact that there are actual courses out there that are effective. There might be one that is better suited to your learning style in particular, but all involve the same basics of memorization.

I for one would not want to try and start out learning English based on puritan works. The complexity would greatly increase the learning curve. Nature teaches us that some processes of learning are the best. Step upon step, line upon line.

 

[TempvsSolvs]

For what it's worth, when I took Latin for 6 years between high school and college, it was heavily focused on 3 areas: first and foremost was grammar (all declensions, conjugations, special cases..), then vocabulary (with a focus on derivative words: English vocabulary played a big role in memorizing Latin vocab...and vice versa, actually), and finally, "culture & civ" where we learned the history and culture, and translated the classics from Pliny, Cato, Cicero etc. for practice with both words and grammar together.

Not to beat a dead horse, since it's been mentioned a few times already, but unlike English which is very contextual, Latin uses very specific constructs for each explicit use of a particular word or sentence structure... Grammar in Latin is essentially a non-negotiable if you want to make progress, especially when you're reading texts of an theological or academic nature. But I suppose using the Reformed writings in place of classical Roman writings may be a suitable substitute for translation practice while getting you closer to your goal.

 

[jhcampbell]

Here is my thing. It is hard for me to believe that Lingua Latina, a book published 70 years ago, is still the best tool we have today, in 2026, for learning Latin. I know that's how some did it, but the reality is, AI does change a lot of things. Figuring out how to implement AI and the older methods to come up with the best current method is the hard part. I guess I will have to research it more.

Part of the issue is that you keep trying to innovate and find the best way instead of going with a way and sticking with it. You’ve been talking about learning Latin for atleast a year now, and as far as I am aware, you have learned NONE.

If you stuck with a book or whatever and actually went with it you’d be much farther along than trying to find the better or more efficient way and not doing it at all.

I’m not saying lingua Latina is the best way I’m saying that it’d be better for you to actually stick with a plan

 

[RamistThomist]

Here is my thing. It is hard for me to believe that Lingua Latina, a book published 70 years ago, is still the best tool we have today, in 2026, for learning Latin. I know that's how some did it, but the reality is, AI does change a lot of things. Figuring out how to implement AI and the older methods to come up with the best current method is the hard part. I guess I will have to research it more.

I never said anything about Lingua being the best book. I should explained how declensions work.

 

[davejonescue]

Part of the issue is that you keep trying to innovate and find the best way instead of going with a way and sticking with it. You’ve been talking about learning Latin for atleast a year now, and as far as I am aware, you have learned NONE.

If you stuck with a book or whatever and actually went with it you’d be much farther along than trying to find the better or more efficient way and not doing it at all.

I’m not saying lingua Latina is the best way I’m saying that it’d be better for you to actually stick with a plan

This is because I don't really care about learning Latin or not. I have 8,000+ English books, and, work on more English books almost every day. What I enjoy doing more than trying to learn something, is attempting to push tech to see if there are ways in which will help the church do things like this easier. If there is no way the tech can expedite the process of becoming a fluent reader in Latin, and instead, someone still has to practice every day for 5-10 years, 1. AI will be so much more advanced in just 2 years, and 2. I can transcribe and check a text in a week or so, which can be auto-translated to a degree of accuracy I will probably never reach. Meaning, with all the English texts out there, and with the way AI can both transcribe (to a high degree of accuracy) and translate (to a high degree of accuracy), there is just no umph to try and personally learn it. The only pleasure I would get, is researching, and trying to find innovative ways to help the next generation in the church learn it (and other languages.) Again, the reason the church needs people to push the tech, is because gate-keeping exists, and in some circles there is an attitude that the old ways are the best simply because they are the old ways. Not calling out people, it is what it is.

I think it just boils down to, I know I can transcribe and translate the books relatively easily. Might as well do that personally, attempt to mitigate errors, and let others who work with languages, be innovative as to give the innovation its credence. It is true, when it comes to languages, it is best to let the people who know them, be the ones to show a "better way."

 

[RamistThomist]

This is because I don't really care about learning Latin or not. I have 8,000+ English books, and, work on more English books almost every day. What I enjoy doing more than trying to learn something, is attempting to push tech to see if there are ways in which will help the church do things like this easier. If there is no way the tech can expedite the process of becoming a fluent reader in Latin, and instead, someone still has to practice every day for 5-10 years, 1. AI will be so much more advanced in just 2 years, and 2. I can transcribe and check a text in a week or so, which can be auto-translated to a degree of accuracy I will probably never reach. Meaning, with all the English texts out there, and with the way AI can both transcribe (to a high degree of accuracy) and translate (to a high degree of accuracy), there is just no umph to try and personally learn it. The only pleasure I would get, is researching, and trying to find innovative ways to help the next generation in the church learn it (and other languages.) Again, the reason the church needs people to push the tech, is because gate-keeping exists, and in some circles there is an attitude that the old ways are the best simply because they are the old ways. Not calling out people, it is what it is.

I think it just boils down to, I know I can transcribe and translate the books relatively easily. Might as well do that personally, attempt to mitigate errors, and let others who work with languages, be innovative as to give the innovation its credence. It is true, when it comes to languages, it is best to let the people who know them, be the ones to show a "better way."

Then you probably need to cut your losses on the Latin front.

 

[davejonescue]

Then you probably need to cut your losses on the Latin front.

True. Its just too easy when I think about with the auto-transcription/translation. I mean if I know it is the long s/f problem where most of the transcription errors occur (within 99% accuracy), I can pin-point those for double checking. If it takes me a week in my spare time, and $15 to transcribe/translate and entire tome, you are right, don't know why I would learn it? It doesn't seem like I would personally be redeeming the time? This way, I only worry about Latin texts if there is something drawing me to them. Its kind of like, alright, do this if need be.

 

[NaphtaliPress]

Again, the reason the church needs people to push the tech, is because gate-keeping exists, and in some circles there is an attitude that the old ways are the best simply because they are the old ways. Not calling out people, it is what it is.

Just stop. Of course you are calling people out; just not by name. Do not bring this up again.

 

[davejonescue]

Just stop. Of course you are calling people out; just not by name. Do not bring this up again.

Sure, no problem.

 

[sc_q_jayce]

I'm interested in how this conversation has transitioned in a way into a discussion of means and ends, particularly in learning.

I was just reading some published papers documenting "metacognitive" laziness due to usage of LLMs and wanted to share a few quotations. Note that both papers are not saying LLM is inherently bad, but I'm just pointing out some interesting observations about the activity of learning in itself. Both papers discuss positive outcomes with LLM but agree in this one particular area.

Our main results are two-fold. First, students in the GPT Tutor (resp., GPT Base) arm perform 127% (resp., 48%) better on the practice problems compared to students in the control arm. This finding is consistent with prior work on the benefits of generative AI on improving human abilities on a variety of tasks. Second, on the exam, students in the GPT Base arm perform statistically significantly worse than students in the control arm by 17%; this negative effect is essentially eradicated in the GPT Tutor arm, though we still do not observe a positive effect. These results suggest that while access to generative AI can improve performance, it can substantially inhibit learning without appropriate guardrails. Importantly, the detrimental impact of GPT Base on learning is of immediate concern since the similar ChatGPT tool is already widely used by students outside of class for help with assignments. Furthermore, an analysis of student interactions shows that students often use GPT Base as a “crutch” by asking for and copying solutions, but they use GPT Tutor in more substantive ways like asking for help or independently attempting answers.‡ Finally, we find evidence that students do not perceive any reduction in their learning or subsequent performance as a consequence of copying solutions, suggesting they are not aware of how generative AI can impede their learning. Our results have significant implications for generative AI tools—while such tools have the potential to improve human performance, they must be deployed with appropriate guardrails when learning is important.​

​

Bastani, H; Bastani, O; Mariman, R, Generative AI without guardrails can harm learning: Evidence from high school mathematics, PNAS, 122 (26), e2422633122

Inversion Effects of AI Undermining Deep Learning​

A major concern associated with an inadequate use or implementation of AI in learning contexts is its potential to undermine the acquisition of knowledge and skills (Huber et al., 2024; Kasneci et al., 2023). This phenomenon is already known from workplace learning, specifically from contexts where AI-driven automation increasingly complements or replaces human actions (Rafner et al., 2021). Whether such effects are problematic depends on the specific conditions and underlying goals. Shifting toward system monitoring may be desirable for enhancing performance and efficiency if AI outperforms humans in a given task. In this context, hybrid intelligence aims to create synergies between human and AI capabilities to optimize problem-solving and task performance (Akata et al., 2020). However, replacing human actions with AI can be problematic if it leads to the loss of human skills critical for achieving high-quality outcomes, diminishing expertise previously acquired and maintained through regular practice. Over-reliance on algorithms may weaken decision-making and judgment skills, resulting in poor choices when AI support is unavailable (Sutton et al., 2018). Especially when AI minimizes human input, professionals may accept its outputs uncritically (Hoff, 2011). This concern is particularly relevant in highly specialized fields like finance and medicine, where professionals bear significant responsibility and their decisions can profoundly impact others (Levy et al., 2019; Mascha & Smedley, 2007).​

​

Bauer, E; Greiff, S; Graesser, AC; Scheiter, K; Sailer, M, Looking Beyond the Hype: Understanding the Effects of AI on Learning, Ed. Psych. Rev., 37 (2025), 45

I've probably used an LLM a total of ten times for work.

 

[davejonescue]

I'm interested in how this conversation has transitioned in a way into a discussion of means and ends, particularly in learning.

I was just reading some published papers documenting "metacognitive" laziness due to usage of LLMs and wanted to share a few quotations. Note that both papers are not saying LLM is inherently bad, but I'm just pointing out some interesting observations about the activity of learning in itself. Both papers discuss positive outcomes with LLM but agree in this one particular area.

Our main results are two-fold. First, students in the GPT Tutor (resp., GPT Base) arm perform 127% (resp., 48%) better on the practice problems compared to students in the control arm. This finding is consistent with prior work on the benefits of generative AI on improving human abilities on a variety of tasks. Second, on the exam, students in the GPT Base arm perform statistically significantly worse than students in the control arm by 17%; this negative effect is essentially eradicated in the GPT Tutor arm, though we still do not observe a positive effect. These results suggest that while access to generative AI can improve performance, it can substantially inhibit learning without appropriate guardrails. Importantly, the detrimental impact of GPT Base on learning is of immediate concern since the similar ChatGPT tool is already widely used by students outside of class for help with assignments. Furthermore, an analysis of student interactions shows that students often use GPT Base as a “crutch” by asking for and copying solutions, but they use GPT Tutor in more substantive ways like asking for help or independently attempting answers.‡ Finally, we find evidence that students do not perceive any reduction in their learning or subsequent performance as a consequence of copying solutions, suggesting they are not aware of how generative AI can impede their learning. Our results have significant implications for generative AI tools—while such tools have the potential to improve human performance, they must be deployed with appropriate guardrails when learning is important.​

​

Bastani, H; Bastani, O; Mariman, R, Generative AI without guardrails can harm learning: Evidence from high school mathematics, PNAS, 122 (26), e2422633122

Inversion Effects of AI Undermining Deep Learning​

A major concern associated with an inadequate use or implementation of AI in learning contexts is its potential to undermine the acquisition of knowledge and skills (Huber et al., 2024; Kasneci et al., 2023). This phenomenon is already known from workplace learning, specifically from contexts where AI-driven automation increasingly complements or replaces human actions (Rafner et al., 2021). Whether such effects are problematic depends on the specific conditions and underlying goals. Shifting toward system monitoring may be desirable for enhancing performance and efficiency if AI outperforms humans in a given task. In this context, hybrid intelligence aims to create synergies between human and AI capabilities to optimize problem-solving and task performance (Akata et al., 2020). However, replacing human actions with AI can be problematic if it leads to the loss of human skills critical for achieving high-quality outcomes, diminishing expertise previously acquired and maintained through regular practice. Over-reliance on algorithms may weaken decision-making and judgment skills, resulting in poor choices when AI support is unavailable (Sutton et al., 2018). Especially when AI minimizes human input, professionals may accept its outputs uncritically (Hoff, 2011). This concern is particularly relevant in highly specialized fields like finance and medicine, where professionals bear significant responsibility and their decisions can profoundly impact others (Levy et al., 2019; Mascha & Smedley, 2007).​

​

Bauer, E; Greiff, S; Graesser, AC; Scheiter, K; Sailer, M, Looking Beyond the Hype: Understanding the Effects of AI on Learning, Ed. Psych. Rev., 37 (2025), 45

I've probably used an LLM a total of ten times for work.

These papers seem to highlight that yes, AI is beneficial for learning; but, the fear is that users may use it as a crutch, so that instead of learning something, in some cases they simply use it for the answers. This then does away with the needs for both traditional instruction and human instructors, while also forming a dependence on the tech. I can see that. I also don't see why people are fighting the inevitable. Across tons of sectors, industries are waking up to the realization AI allows a single one of their workers, many times to do the work of 10 people. So that these industries are not void of human interaction, it can just be minimized in such a way that saves these industries tons of money. I don't see that odd if the goal of capitalism is the best ROI? Will people use AI as a crutch? Absolutely. The same way people use calculators, Google, and Wikipedia. But since AI is more advanced than all of these, its touch and effects will pour over into more areas. It is what it is. There is no stopping it. With this in mind, the goal probably shouldn't be so much in trying to dissuade people from using AI, though that may work for older generations, it will be unheard of for the ones coming up; instead people should focus on trying to see the trajectory of AI usage, and attempt to implement best practices.

This is why I was saying for me. Why am I going to spend 10 years trying to learn Latin, if I can transcribe/translate a document and have it converted to a good degree of fluency and accuracy, I will probably never reach trying to learn it? All for the cost of a Sonic combo meal? It doesn't make sense to me to spend all those hours attempting to learn it, when, like the calculator for advanced math, AI makes doing so unneeded. For me, if AI could expedite the process, then sure, I may give it a shot; but, if like others are saying, it cant, then there is little reason to. And again, this is me personally, but I wouldn't find any glory in learning Latin in itself, it would be just in an attempt to read these works. I have already had feedback from former AI generated translations, that they are pretty/reasonably good. Compare that to how long it would take a human to reach the same level of proficiency in that language, go through a text, and produce something, "how much better?" I want to say Charles was saying he spent roughly 2-3 years doing a single book (out of 10) of the Syntagma, and, he is proficient/expert at Latin. They are going on 10 years for Mastrichts Theo-Practical. And this is from Latin experts, people who are the top of their fields. Brother Dilday, I think has been working on the books he is doing for over a decade now. Kind of see what I am getting at? That is, what degree are the proficient/professional translations (outside of elegance) that much better, to struggle so long to learn it, to take so long to translate something (in the face of the now present alternative?) Again, for me, if AI cannot expedite the learning process, it seems not only easier, but more expedient simply to use AI like a Latin calculator, and have it do the work. Why struggle for 10 years, when I can do a batch API and have it translated in its entirety in 10-30 minutes?

If that is a crutch, so be it. But, that crutch will possibly allow the reading of works that 1. Will never get translated in my lifetime, and 2. Few ever get the chance to read because of that. At this point though, all this/that is beating a dead horse. The end of the matter is simple: Some people do not agree nor desire to use AI in such a way, and some people do agree and will use AI in such a way. Let each do what they are going to do.

 

[RamistThomist]

Why am I going to spend 10 years trying to learn Latin, if I can transcribe/translate a document and have it converted to a good degree of fluency and accuracy, I will probably never reach trying to learn it?

Because knowledge transforms. Embarking on such a linguistic project changes you.

 

[davejonescue]

Because knowledge transforms. Embarking on such a linguistic project changes you.

I agree. In no ways am I saying there is no value in that, there absolutely is; the same way diet and exercise will change you when losing weight, as compared to Ozempic. I am only saying for me, that value is not worth it. I am not concerned with the changing, I am concerned with the reading. If God desires to teach me discipline, he can do it in a variety of different ways. God has allowed AI to enable such translations. God has enabled me to be able to learn how to utilize AI to produce them. So when discipline is concerned, it will most likely not be a language he uses if/when he attempts to teach me discipline.

 

[RamistThomist]

I agree. In no ways am I saying there is no value in that, there absolutely is; the same way diet and exercise will change you when losing weight, as compared to Ozempic. I am only saying for me, that value is not worth it. I am not concerned with the changing, I am concerned with the reading. If God desires to teach me discipline, he can do it in a variety of different ways. God has allowed AI to enable such translations. God has enabled me to be able to learn how to utilize AI to produce them. So when discipline is concerned, it will most likely not be a language he uses if/when he attempts to teach me discipline.

I get the cost-benefits analysis. I could read and (relatively) speak German 12 years ago. I could probably do it again, but I am not going to invest the energy for it. On the other hand, learning Latin (or Greek) makes your brain work in a different way. It's a qualitative step above just "teaching discipline."

 

[davejonescue]

I get the cost-benefits analysis. I could read and (relatively) speak German 12 years ago. I could probably do it again, but I am not going to invest the energy for it. On the other hand, learning Latin (or Greek) makes your brain work in a different way. It's a qualitative step above just "teaching discipline."

Nice. That is going to be so awesome for those that do it. In a few short years, in glory, either my brain will work exactly the way God intended (meaning, applying the benefits learning it would render), or, I will have an eternity to learn Latin. Either way, its not a big deal to me now, with the added bonus of AI allowing relatively quick and inexpensive access to these works. If we lived in a world where AI didn't exist, and there were no relatively quick or inexpensive access to these works, I may feel more pressured to learn it. But we do live in a world where it does exist, so I don't feel as pressured.

I tried looking for ways or suggestions in which the present tech may expedite that learning. According to the consensus here, there isn't any. So, there seems no reason to me to bother. Again, when I look at the contrast, i.e. 10 years vs 10 minutes to a text, I will stick to the latter for the limited works I would like to read, even if that means missing out on the other benefits learning a language may offer. This is just me. Others of course are free to do as they desire.

 

[MW]

I can't imagine what it would be like for someone to learn English by extracting vocabulary from a theological text book. It makes much more sense to begin with the basics of real world expressions and build one block on top of another. Polanus, Voetius, et al. learned Latin in this mechanical way. They built up their knowledge base over years of practice. And then what they learned was expressed intuitively in their writing in complex ways.

 

[davejonescue]

I can't imagine what it would be like for someone to learn English by extracting vocabulary from a theological text book. It makes much more sense to begin with the basics of real world expressions and build one block on top of another. Polanus, Voetius, et al. learned Latin in this mechanical way. They built up their knowledge base over years of practice. And then what they learned was expressed intuitively in their writing in complex ways.

Yes, I agree. With only one caveat, that is they most likely (like I think many did back then) start learning Latin at a very young age. I am 44. If there is no way to expedite the process, given the present tech, it seems natural to use the same tech to bypass it (if it is able.) That is, with allowing 30 minutes, to an hour a day, 4-5 days a week; how many years would it take me to be able to read something like Polanus or Voetius fluently and enjoyably? I cant speak to your Latin fluency, or if you have tried to translate texts on here before; but it looks like you are doing so now that AI can expedite that process. Which means even with all the years you have been dealing with it (if you have) where you capable of dealing with it as you do now? If not, its such examples why I see little reason, as a middle-age man, to start something that can pretty much entirely be bypassed. If for instance the traditional means even leads those who learned by it, after all that struggle, to eventually utilize AI in bulk translation? Its no shade, I think it fully capable, and a lot of the people on Fiverr who do Latin translation make it upfront that AI assists. I really don't see it as novel or trivial anymore. I looks like the way it will mostly go from here on out.

I think if we are being honest, not a lot is changing. Very few Protestants know Latin fluently. Much less than those who know Greek and Hebrew fluently, and that is few. I don't think something like AI can tarnish an almost already non-existent venture, by reducing those who would seek to learn it. All it is doing is allowing not knowing it, no longer to be an obstacle to reading some of these old works. Thats it.