Why don’t y’all just get to machine learn all those fucking tablets you dug up, like hundreds of thousands of them, and train a fucking AI on that shit and tell us what it says instead of sitting here being a besserwisser online, HMM? If there was one good cause for AI, cuneiform would be it. Just god damned saying.
Edit: just btw I happen to know that the problem is mainly the first training set, you need cuneiformers to correctly give the answers so the model knows what to train on, and there’s like seven people in the world who do that, but I’m thinking, what if we trained an AI model on all the cuneiform we do know? Hit me up for proposals, I’m serious about this shit
First off I don’t study ancient Mesopotamia, I just have actual historical training so I know how to look at a piece of evidence and be able to see how it can be interpreted in multiple different ways.
Second, machine learning is not a cure all solution. The time that it will take to get a machine learning algorithm to even be half of a human translator can do is just wasteful of our time. Machine learning will produce a ton of errors that will have to be dug through by those some limited translators. Meaning they won’t be doing what we actually need them to do for prolonged periods of time.
Third, even if (and that’s a big if) it could figure out translations for things we haven’t been able to, we would still have a hard time verifying it’s accuracy till translate it ourselves. We are dealing with a fully extinct language here, we had to decipher what it actually means. We are dealing with fragmentary artifacts too, something which no AI model could accurately fill in the blanks of.
Fourth, cool call me names because I specialized into a different field. I’m going to leave the actual translation efforts to the experts in ancient Mesopotamia, you should too.
Why don’t y’all just get to machine learn all those fucking tablets you dug up, like hundreds of thousands of them, and train a fucking AI on that shit and tell us what it says instead of sitting here being a besserwisser online, HMM? If there was one good cause for AI, cuneiform would be it. Just god damned saying.
Edit: just btw I happen to know that the problem is mainly the first training set, you need cuneiformers to correctly give the answers so the model knows what to train on, and there’s like seven people in the world who do that, but I’m thinking, what if we trained an AI model on all the cuneiform we do know? Hit me up for proposals, I’m serious about this shit
First off I don’t study ancient Mesopotamia, I just have actual historical training so I know how to look at a piece of evidence and be able to see how it can be interpreted in multiple different ways.
Second, machine learning is not a cure all solution. The time that it will take to get a machine learning algorithm to even be half of a human translator can do is just wasteful of our time. Machine learning will produce a ton of errors that will have to be dug through by those some limited translators. Meaning they won’t be doing what we actually need them to do for prolonged periods of time.
Third, even if (and that’s a big if) it could figure out translations for things we haven’t been able to, we would still have a hard time verifying it’s accuracy till translate it ourselves. We are dealing with a fully extinct language here, we had to decipher what it actually means. We are dealing with fragmentary artifacts too, something which no AI model could accurately fill in the blanks of.
Fourth, cool call me names because I specialized into a different field. I’m going to leave the actual translation efforts to the experts in ancient Mesopotamia, you should too.