In my experience as a translator, the utility of these large language models is highly variable depending on several factors. These include:

• The particular language pairing

• The direction of translation within that pairing

• Whether the source text is closer to the written or spoken end of the spectrum of communication mode

• The types of information and ideas being conveyed

• Whether the text involves formulaic or non-formulaic syntax and semantics

Anyone who believes that LLMs—comically misnamed as “artificial intelligence”—can be applied more or less uniformly to provide solutions to all translation tasks is profoundly ignorant about what LLMs actually are and how they work.