Show HN: Bible translated using LLMs from source Greek and Hebrew

Yes agree on translation. The story of arriving at a word is usually more interesting than the word itself.

In an ideal world we could ingest the full study Bible's notes. My guess is much of the NET-level (or other study bible) scholarship is part of the base model corpus.

Here's a deeper view into the verse process. We did use a reasoning model.

{ "reference": "Genesis.1.1", "optimal": "In the beginning God created the heavens and the earth.", "poetic_daily": "At the dawn of all things, God shaped sky and soil into being.", "footnotes": [ { "anchor": "God", "note": "Hebrew אֱלֹהִים (Elohim); plural in form but singular in meaning when referring to Israel’s God." } ], "controversies": [], "connectives_check": [ { "source": "וְ", "rendered": "and", "status": "kept" } ], "consistency_flags": [], "scholars": [ { "name": "Thomas Schreiner", "one_sentence_view": "Emphasizes the verse as the absolute beginning of creation, affirming God's sovereign initiative." }, { "name": "Walter Brueggemann", "one_sentence_view": "Sees the verse as a theological overture introducing God’s ordering power over chaos." }, { "name": "Eugene H. Peterson", "one_sentence_view": "Views the line as the opening note of a grand narrative, inviting readers into God’s creative story." } ], "lexeme_refs": [ { "anchor": "God", "lang": "he", "lemma_id": "430", "note": "אֱלֹהִים (Elohim) chosen per rule preference; conveys the singular Creator without plural nuance." }, { "anchor": "and", "lang": "he", "lemma_id": "c/853", "note": "Connective וְ/ marks coordination between 'the heavens' and 'the earth'; retained explicitly." } ], "review": {} }

100% agree, like any technology it's neither inherently good nor evil.

In this instance, I think it has the opportunity to democratize deep religious study in ways that used to be reserved for serious scholars.

e.g. Do you know what the word "daily" in the Lord's Prayer comes from?

Questions like these can engage the mind and spirit.

I hope more people use the tools to fully explore their faith, instead of outsourcing prayer and sermon creation to the LLMs.

You read multiple English translations and you get your own sense of how biblical text can and is rendered.

Admittedly it's an aesthetic judgment on certain verses.

There are a great many views about this depending on who you talk to. In Christian circles, it’s essentially the infallibility vs inerrancy topic, with fundamentalist denominations leaning toward inerrancy (which is the view that original manuscripts have complete historical accuracy).

Obviously, you have to take a strong “religion first” lens to everything about the world from there.

But of course, there were ancient cultures that pre-date Judaism (and by extension Judeo-Christian sources), which share many similar stories but with different details and descriptions. Large scale flood myths and arks are common in history. You can read the Mesopotamian version in the Epic of Gilgamesh, which is strikingly similar to Noah’s ark.

Different models only. Same process.

There is and will be more details on process in future blog posts (blog is at the very bottom)

But the Peshitta is 300 years after the Septuagint and the verse you mention was written in Greek in a gospel, not the Hebrew Bible. I don't know how the translators of the Peshitta would have any special access to sayings of Jesus that predate the Gospels we have now. I don't know if there is any hard evidence that the Gospel authors had actual eye witness written sources in the language that Jesus spoke. So you have to assume that Jesus used Gamla in Aramaic and that the Gospel writer mistranslated it when writing in Greek but that the Peshitta gives special insight by retranslating it back to the ambiguous word. Then you have to make another leap to think that it is somehow possible to manipulate a rope to be the size of a thread. Sounds like a lot of histrionics to justify Abraham going to Paradise when the simpler explanation is that it's just a concept of difficulty rather than a logical word problem. This seems less plausible than the Eye of Needle gate theory which many Christian teachers often reference.

> Heck, there's others who were wealthy in scripture, even kings are they all doomed?

This is a great question. In the next verses, the disciples ask pretty much the same thing: "Who then can be saved?" and then Jesus explains to them:

    With men this is impossible, but with God all things are possible.

Whether it's a camel or a rope (and whether it's a literal needle or a small city gate, as some people argue), I think is less important (though still interesting). Either way, after the rich young ruler walks away, Jesus turns to his disciples and paints a picture something that's completely impossible without God, no matter how hard we might try by ourselves.

[flagged]

This is definitely one area where the training set for the LLM is liable to be polluted by existing translations and even straight memorized english biblical text.

Not to be too much of a devil's advocate (ha!), but I kindof think I _want_ it to have biblical translation data in the dataset. So long as it's not simply copying something like the KJV or ESV into the output, then this should be a good thing, right?

Because much of what it produces (especially in the "poetic" mode) does seem to be very much "off the beaten path" for a good number of renditions.

I don't think that the goal would be to have a dataset that is completely free from scholarship on the topic of Biblical translation, but rather to synthesize the rules and principles from the collected body of knowledge and apply it (with steering) to the entire Biblical text.

Thoughtful critique.

No one is suggesting you replace your ESV or NKJV with this for your religious study. This is as much a technical project of interest as it is a faith-based one.

In terms of your view of the priors on the Bible, you've described in my experience the process all translations go through. We're all skewed by default toward reproducing (poorly) previous translation through word choice modification.

That is, in many ways, the whole thing. My guess is an iterative approach can actually yield a better approach as words shift meaning socially over time.

But we will see!

Weren't some pretty critical points of the Bible, Hallucinated originally? What's the problem?

[flagged]

Thanks for noticing. The blog has more detail; these are first iterations and the verse and editorial notes sometimes slip in :-(

Side project, big book.

Those will get ironed out soon.

Not here to have a religious debate - though given HN, it may turn into one!

Imprecise language is a common human feature of a lack of understanding - something we all suffer from. We call LLM's "AI" without fully understanding what's artificial and what's intelligence.

The story of faith is, in some ways, the story about how little we know about the universe. That doesn't mean there's no progress. If anything, it shows there is an end goal.

The ancient narratives of Babel and Genesis reduce the incomprehensible (Creation, the Divine) into elements we as humans at that time could understand.

How else could our ancestors have possibly related to the divine?

I mean, I should hope so? I don't think the point is to create a completely "from scratch" translation -- this feels like a more practical exercise than academic.

What's fascinating to me is the suitability of some of the renditions that have been made. I've done a little bit of spot-checking, but some of the renderings (especially in the poetic version) don't seem to be very common translations.

Ideally this would be building off of established practice for translation, synthesizing existing human work, and applying those aggregated principles over the entire corpus.

It doesn't seem to simply be going with the majority version.

For the moment, no. But it will be in the next few months.

There are lingering errors (hallucinations, notes) that need to be removed before it would be reasonable sharing and studying broadly.

The other commenter has pointed out novel copyright issues, not that it's likely we'd go to court over this.

A complicated question. They assert copyright, but whether or not that would hold up in a court of law in any given jurisdiction is still largely an open question.

The main page? It's a static site on cloudflare - the LLM calls are openrouter and are rate limited so that might be what's going on.

If you have more details, we'll investigate.

It was slow to load for me but loaded eventually.

Say more - why?

Agree.

And if we could bring more faithful people into that agreement process, that's a good thing.

As for the personal achievement, nothing really is a fully personal achievement in this (or really any) domain.