I've seen a non-native speaker trusting an AI translation tool to draft the German version of a minor document change.
-
RE: https://mastodon.social/@hanno/116193243981099211
I've seen a non-native speaker trusting an AI translation tool to draft the German version of a minor document change.
The AI tool hallucinated a subtle mistake in the written-out number - "63872 (in Worten: Dreiundsechzigtausendachthundertundsiebzig)" - an error which the non-native speaker was obviously unable to spot.
Translation is often hailed as a positive example of using AI, but it suffers the same problem of human nature to accept a machine's plausibly looking result as "good enough" without the review that would be required.
-
RE: https://mastodon.social/@hanno/116193243981099211
I've seen a non-native speaker trusting an AI translation tool to draft the German version of a minor document change.
The AI tool hallucinated a subtle mistake in the written-out number - "63872 (in Worten: Dreiundsechzigtausendachthundertundsiebzig)" - an error which the non-native speaker was obviously unable to spot.
Translation is often hailed as a positive example of using AI, but it suffers the same problem of human nature to accept a machine's plausibly looking result as "good enough" without the review that would be required.
@hzulla Related to this… the ongoing challenges at Wikipedia with people using LLMs for translation of articles… and wind up having false info inserted into the translated articles: https://www.404media.co/ai-translations-are-adding-hallucinations-to-wikipedia-articles/
-
E em0nm4stodon@infosec.exchange shared this topic
