Which is great until it blithely translates an ancient curse and initiates the apocalypse.
Given how chaotic people are, it's remarkable that millions of us working together isn't invariably a constant raging dumpster fire, and instead is only a bit of smoke here and there that sometimes flares up a bit.
It's an organelle that periodically contracts to expel water from inside the cell, out. This prevents the cell from absorbing too much water through osmosis and rupturing.
Even the individual cells are bailing water, as if from a sinking life raft. Always have been. Life itself is a process of dysfunctional function. Actively resisting death. It makes sense that we see this reflected at all scales. It is not a disorder, it is the natural state of living in a universe with entropy.
Can machines curse?
Also, "acquired" could simply mean that the museum paid money to a third party to, well, acquire the piece(s) in question while not inquiring too deeply into how said third party came into possession of the objects that they sold.
No problem marking them as machine translated and keeping track of which have been spot-checked by experts either.
Indeed, one of the thing you'd probably like the translators to do is identify rare or unique words that can be added to our existing knowledge of these languages.
It would be really neat to set up something like a wiki populated with the existing translations and machine translations done via LLM, and to periodically re-train the LLM on all the newly manually verified translations and automatically re-run the machine translations after. The whole thing could move incrementally toward high quality output.
(I think I just said "triage" in a lot more words, to be honest).
They got photos to the leading professor in Ogham who translated it for them.
I happened to mention this on HN a few months ago in relation to something or other, and a commenter replied to me to explain that that translation was no longer sound, and the current understanding was that the tablet said something completely different instead!
(Found the link they gave! https://www.babelstone.co.uk/Blog/2008/05/throng-of-fifty-wa...)
So the done thing in translation is to choose some existing translation as a "gold standard" and use that one, without any assumptions of how good or bad it is. Sometimes there's an attempt to rate translations by polling humans but that can not be easily done at scale and certainly not at web-scale, as in LLMs (the modern de facto standard for automatic translation as far as I can tell; I haven't been watching the field closely lately).
The same logic is applied to metrics like BLEU and ROUGE scores, used to measure the goodness of a translation. The general idea is that you choose a translation to be the "gold standard" and then compare the n-grams in the gold standard, and the automatic translation, for overlap.
It's a very crude and imperfect measure and it's one reason why, in practice, we have no idea how good automatic translations are especially now that automatic translation systems are deployed and working every day with millions of input and output texts that nobody can reasonably be expected to evaluate.
In any case just because there's no fixed opinion on what is a good translation, doesn't mean an LLM can't be used to produce one. It probably will. Assuming I understand your point better now, I agree that this is going to cause trouble down the line.
Given what and where they were from it almost certainly means donated or sold to the museum by the archeologists who excavated them in the first place.
"From 1877 to 1882, while undertaking four expeditions on behalf of the British Museum, Rassam made some important discoveries. Numerous finds of significance were transported to the museum, thanks to an agreement made with the Ottoman Sultan by Rassam's old colleague Austen Henry Layard, now Ambassador at Constantinople, allowing Rassam to return and continue their earlier excavations and to 'pack and dispatch to England any antiquities [he] found ... provided, however, there were no duplicates.' A representative of the Sultan was instructed to be present at the dig to examine the objects as they were uncovered."
So, not a euphemism for "stolen".
In Eurasia and Africa this does not work. "The oldest traceable group" may be many invasions, conquests and mass migrations after the things were made. The current residents or earliest traceable group may well have enslaved, oppressed, or killed the group who made the things. In many, many cases they would have erased the culture that made them.
In the case of these tablets there is no one who can meaningfully claim to be an Akkadian.
With a few minutes of searching the internet, you would not have written that inappropriate question.
Tablets are 3D objects which are most often eroded and broken. A plain photography, like one would make for a sheet of paper, is useless in the general case. To bring out the cuneiform characters, the light must be raking the surface. And don't forget that that surface can be concave, and text can go over the sides. Most tablets need many photos with different lights. It's a long and hard work that is not automated.
Guess what, Scholars have heard about AI. If AI could help them publish astonishing papers and push forward their career, do you really think they wouldn't rush to it?
Look at https://etcsl.orinst.ox.ac.uk/cgi-bin/etcsl.cgi?text=t.3.1.2... , for example. To me it doesn't come across as something I would expect an LLM to be able to produce. They are basically decent at translating between certain languages, and only documents that are very formal. As soon as metaphor, idioms and so on come up in source texts they suck.
There's more translated sumerian you can read here: https://etcsl.orinst.ox.ac.uk/edition2/etcslbycat.php
This would be an awesome job for AI. Get these scholars to train the AI, then have it automatically translate all the other tablets
Also, are there online scans of the tablets? Maybe the problem is not the scarcity of scholars, but gatekeeping?
On a tangent:
> These tablets, which likely come from Sippar, a Babylonian city in modern Iraq, were acquired by the museum more than a century ago and date to between 1900 and 1600 B.C.
Is "acquired" a euphemism for stolen here?