Can AI solve the problem of importing TeX/LaTeX into TeXmacs?

[amichail]

Although TeX is Turing-complete, perhaps AI can enable high-quality, ever-improving imports of TeX/LaTeX from your coauthors?

[schroedingers_possum]

Hi,
I’m not completely sure if I understand correctly, so please correct me if I’m wrong.

I assume that you are talking about using a large language model to convert LaTeX into the TeXmacs format.

From my experience, the latex import already works really well and things that don’t convert well often have more to do with some packages that aren’t supported in TeXmacs, so I doubt that using a LLM would help in this case.

Theoretically, you could probably try using a LLM to first convert LaTeX segments that have trouble converting to only use standard packages (e.g. a friend of mine likes to use an esoteric physics package for bra/ket and vectors instead of amsmath).

Do keep in mind that LLms can sometimes change the content of the text, so manual proofreading would be required whether you make use of LLMs, or not.

[Jayson]

Looking for something like this?
@darcy 's team is still working on importing TeX and Markdown.