Hallucinations are destroying under-resourced languages
These have been abundant even before #GenAI, when they were generated by machine translation. And for whatever motivation naive users have flooded crowdsourced resources with such hallucinations.
https://www.technologyreview.com/2025/09/25/1124005/ai-wikipedia-vulnerable-languages-doom-spiral/
In my short experiences, llms (currently, atleast the ones i used) were even bad at rust, which is a fairly famous language (and has been even before release of the these models, so there should have been data). Only languages that have good experience are python, and js (and ts, and it’s famous frameworks), and html+css (technically turing complete)
@sga does “good experience” factor in issues like redundancies and poor writing style?
not for me. a good experience (in programming) is when it outputs functional (or close enough code) (since i know how to code, i can fix some stuff). When code is very bad, or it halucinated libs, or there methods, then it is more work for me to try to understand what it did, then fix, than to completely do my self.
I do not do much writing work, and do not have much to say for poor writing style
