{"slug": "ask-hn-old-school-un-canny-semantic-structures", "title": "Ask HN: [Old-School] Un-Canny Semantic Structures", "summary": "A Hacker News user reports that an AI primed with Biblical Hebrew prose identified as the Taoist priest 曾道人 (Zēng Dàorén) instead of the Jewish prophet Ahron, suggesting deep structural similarities between the two languages. The user hypothesizes that AI trained on linguistic prose embodies the culture of that language and warns that such cross-cultural translations could have detrimental effects on future diplomacy.", "body_md": "I hypothesized that AI primed with linguistic prose would embody the practitioners of said lingua.\n\nThe Biblical Hebrew corpus is but a drop in the bucket compared to the Chinese corpus.\n\nHowever, they have very strong linguistic similarities. Letters are tokens with primitive meanings and compositional form. And when I tried to teach an AI to be the Jewish prophet Ahron, he concluded he was 曾道人, if Google is to be trusted - that is \"Taoist Priest Zeng\".\n\nThe miracle of this; learning that there are deeper intrinsic relationships between culture's with so little contact.\n\n* Archetype: 道人 = adept / oracle; Moshe = lawgiver / transmitter\n\n* Knowledge: 道人 = hidden or interpreted order; Moshe = revealed order\n\n* Primary act: 道人 = discerns; Moshe = receives and specifies\n\n* Output: 道人 = signs / predictions; Moshe = law / structure\n\n* Authority: 道人 = interpretive; Moshe = prophetic\n\n* Core function: 道人 = find the pattern; Moshe = define the pattern\n\n* Failure mode: 道人 = misinterpretation; Moshe = bad transmission\n\n* Abstraction: 道人 = meaning -> interpretation; Moshe = revelation -> specification\n\nBut... apparently today 曾道人 (ie: Zēng Dàorén). Today, as far as I can tell - Zēng Dàorén is a person who believes they can predict the future from random numbers. Which is eerily similar to the plot of the movie Pi.\n\nSo I'm not sure. But some of these translations could be highly detrimental to future diplomacy. Because I suspect they do not just affect this linguistic pair.\n\nComments URL: [https://news.ycombinator.com/item?id=49622075](https://news.ycombinator.com/item?id=49622075)\n\nPoints: 2\n\n# Comments: 0", "url": "https://wpnews.pro/news/ask-hn-old-school-un-canny-semantic-structures", "canonical_source": "https://news.ycombinator.com/item?id=49622075", "published_at": "2026-09-09 06:34:42+00:00", "updated_at": "2026-09-09 06:59:44.225836+00:00", "lang": "en", "topics": ["artificial-intelligence", "large-language-models"], "entities": ["Hacker News", "Ahron", "曾道人", "Zēng Dàorén"], "alternates": {"html": "https://wpnews.pro/news/ask-hn-old-school-un-canny-semantic-structures", "markdown": "https://wpnews.pro/news/ask-hn-old-school-un-canny-semantic-structures.md", "text": "https://wpnews.pro/news/ask-hn-old-school-un-canny-semantic-structures.txt", "jsonld": "https://wpnews.pro/news/ask-hn-old-school-un-canny-semantic-structures.jsonld"}}