Comment Re: Tip of the iceberg. (Score 1) 55
Yeah, but I don't really blame people. LLMs work so incredibly well and produce amazing results. Even AI experts are impressed by the results and non-AI experts are often seduced by the results. Well, except for the anti-AI folks who, as usual, see all of the warts and none of the benefits, and are thus exactly as idiotic as the all-AI-all-the-time folks, just in the opposite way.
In truth, AI does mostly-great work and will replace many jobs over the next decade; translation jobs are very high on that list. But they need systems to double-check them, some computer and some human, because the mistakes AI makes are often hilariously bad (hallucinating relevant cases in legal filings) and often subtle and hard to detect. People want to replace people but expect AI to not need validation, even though most all-human systems have (and need) validation.