GPT is way better at next-word predictions than n-grams. However, I agree with OP that the resulting compression scheme is probably not very practical all things considered.
Oh yes, thank you for saying that. We seem to be the only ones here that don't like this. I have a 2K screen, please let me see more content, not less on it.
You could use a static website generator such as Jekyll or Hugo. Then, if the tools stop working for any reason, you always have the generated HTML than you can update.
I agree. The other day, I had to write a 1-page project proposal for a PhD scholarship and they were giving the font to use and its size, and the margins. I had almost no mathematical symbol to use. One of the professors said that it looks unprofessional to not use Latex. I mean, I could have use Latex and I would have looked like exactly the same given the requirements. I just did it in LibreOffice because that was really the simplest I could have done.
Almost everything said on that page is already feasible with basic network spoofing. Sure, the hardware hack is cool and allows a bit more but nothing extraordinary.
> it is better to rely on ML than on low-quality theory.
Isn't ML (and especially deep learning i.e. neural networks) itself a low-quality theory? From what I understand, nobody really knows why it works so well but uses it anyway.
> There is nothing about: part of speech categories, relative clauses, morphology, affixes or compound words, the theta criterion, the content/function distinction, verb tenses, agreement, or anything else related to actual linguistic phenomena.
When you learned you native tongue, you didn't need to know all of these. You just learned. So, maybe the problem IS about math and algorithms instead of linguistics.