Hacker News Reader

Comment by 0xffff2 | Hacker News Reader

I really hate that I'm even typing these words right now, but wouldn't this be one of the few fields where modern LLMs would potentially provide a real capability that was previously lacking? I'm imagining an LLM might be able to provide rough first pass translations and provide hints about which specific tablets might be most interesting for a human that can't possibly look at them all.

View on HN

There's inevitably been quite a bit of research in using LLMs to help decipher cuneiform, but it's still at a fairly early stage. Good overview here:

https://www.csmc.uni-hamburg.de/publications/mesopotamia/202...

"Artificial intelligence can therefore provide researchers with a significant support, helping them link together texts that mention the same individual, a geographical site, or a word within a large digitised corpus. But we are still far from machine decipherment of texts."

I doubt there's much Akkadian and the 14 other languages written in cuneiform over more than 2,000 years. Several of these are language isolates. Its probably not intractable but would require a lot of human labeling and correcting by academics who can read these languages. It would be a fun legacy project for an eccentric billionaire. If I were Bezos, it would be on my todo list.

and imo ive also thought the same thing, but then it does still come down to the amount of people and time it would take to go over those rough translations and pull the wanted info out of it. manpower is lacking in a lot of those fields

Why would you hate that you're typing those words? Technology is useful and neutral, and if we have good uses for it like it seems so here, let's do it! :)

because it's a fad-ish response right now.

A decade ago every "omg just use [answer]!" was to suggest blockchain solve your problem.

Every possible problem seemed to have a solution that involved blockchain.

... now that "[answer]" is always "AI".

So I get his sense of cringe... while also acknowledging I had the same thought. We've never been better at pattern matching, pattern identification, OCR, translating across languages... seems like a great use of an LLM.

Reply on news.ycombinator.com