

“Just tell it to not make mistakes.”
Yeah, right.


“Just tell it to not make mistakes.”
Yeah, right.
The process for this is usually like that:
Friend of mine applied for a job where they asked for at least 5 years of experience with Angular version x.y.z (can’t remember the exact version). The friend responded that he had 10 years of experience with versions x-3 to x+1.
The HR person doing the hiring asked back “But do you have 5 years of experience with the exact version x.y.z?” to which he answered “Version x.y.z has only been out for 3 years so it’s impossible to have 5 years of experience with it.” HR wrote back saying that he was rejected because he didn’t have 5 years of experience of experience with that exact version.


Sadly we had that problem before AI too… “Some dude I know told me this is super easy to do”
There’s a huge difference between “Creates intelligible single-use text that’s good enough that I can understand what the text is roughly about” and “Creates text at a quality high enough to work as a quotable source”.
For the first use case, infrequent hallucinations are no problem. I read it, if I understand a bit about the topic I might catch it, if not it probably doesn’t matter too much either. Especially if it’s about non-critical topics.
For the second use case, infrequent hallucinations are a massive problem. Most people who use Wikipedia use it like a primary source. Even though sources are linked, they don’t go hunting for sources but instead rely that the information in the article is accurate. Every article is read not only once by one person, but thousands or hundreds of thousands of times. That means every single line is read and believed. You can bet that if there’s a hallucination in there, someone will read it and believe it. That’s requires a completely different level of accuracy, and doing that kind of crap translation work on such a large scale as OKA is a massive disservice.