{"id":69449,"date":"2024-09-30T04:44:09","date_gmt":"2024-09-30T04:44:09","guid":{"rendered":"https:\/\/clearglobal.org\/?p=69449"},"modified":"2024-10-01T05:03:16","modified_gmt":"2024-10-01T05:03:16","slug":"the-challenge-of-language-inclusion-in-the-age-of-ai","status":"publish","type":"post","link":"https:\/\/clearglobal.org\/the-challenge-of-language-inclusion-in-the-age-of-ai\/","title":{"rendered":"The challenge of language inclusion in the age of AI"},"content":{"rendered":"\n

Earlier this month, The Atlantic CEO Nicholas Thompson posted a short video<\/a> on LinkedIn (he does them every day – and they\u2019re super interesting) about this paper,<\/a> saying that Large Language Models (LLMs) are \u201cincreasing the number of people employed in translation.\u201d That\u2019s not exactly what the video said and it is not what the research paper says, but, hey, it made me click on it. <\/p>\n\n\n\n

LLMs, and generative AI models use algorithms trained on a lot of text (A LOT of text) to recognize complex patterns. The more text and greater computing power they employ, the better the model performs. At the foundational level, they learn to respond to user requests with relevant, in-context content written in human language \u2013 the kind of words and syntax people use during ordinary conversation.<\/p>\n\n\n\n

The paper said that LLMs can speed up and even improve translations, particularly for less experienced translators. The researcher looked at training compute, <\/em>which is the amount of computational power required to train an LLM. The more training compute an LLM has, the faster and better a translator can work. In theory, this could lead to translators working more quickly and thus earning more money for the same number of words. <\/p>\n\n\n\n

Clearly, neither Nick Thompson nor the paper\u2019s author work in translation. Post machine translation editing fees are often much lower than fully human translation ones. So, I think it\u2019s a leap to say that more people are making money from translating, but it is reasonable to conclude that translators can work more efficiently.  <\/p>\n\n\n\n

However, I think the research could yield interesting findings about the quality of LLMs and translations in lower-resource languages. The researcher used translators working from English into Arabic, Hindi, and Spanish. It\u2019s well documented that LLMs currently work best in English <\/a>and other languages that have a large digital presence. Spanish also has a large digital presence, but Arabic and Hindi – not so much.<\/p>\n\n\n\n

<\/div>\n\n\n\n

LLMs may make translation slower in some cases<\/strong><\/h3>\n\n\n\n

The study focused on strengthening the argument that the more resources an LLM has in terms of its training compute, the faster and better the translation is. The data supports this hypothesis. It\u2019s hardly surprising. <\/p>\n\n\n\n

The researcher didn\u2019t provide the data on whether or not LLMs helped or hindered translation in the first place. Machine translation (a different application of AI) can speed up translation in languages for which lots of digital data is available, and it\u2019s widely assumed that LLMs are the same, if not better. There has been little (if any?) analysis if the same is true for low-digital-resource languages. And this is the first study I\u2019ve seen where the data suggests that it might take longer <\/em>to translate content using an LLM for Arabic.<\/p>\n\n\n\n

And that\u2019s interesting because it supports the idea of how critical digital language data is. <\/p>\n\n\n\n

There are 2 big issues here:<\/p>\n\n\n\n