YurasApr 11
Jon Sterling

The issue with AI-writing is not related to disclosure, or lack thereof. It's about values, and understanding what the social function of writing is.

That's why university policies requiring "disclosure" of AI assistance in writing are bullshit: if you used it to help you think, you shouldn't have to disclose it, and if you used it to help you write, you shouldn't ever show that writing to another human being. Ever.

Apr 11 at 6:46pmWeb
Show threadDima Pasechnik 🇺🇦 🇳🇱Apr 11

@jonmsterling it's a bit tricky to find the border between "using a good spellchecker" (spell-checkers are everywhere nowadays) and "using AI assistance wrongly".

Or, say, I recently was reprimanded here for boosting a post from @zbMATH where they mentioned use of an LLM-enchanced tool to run OCR on their old reviews and convert them into LaTeX.

Show threadNo mother, no bride, no kingApr 11
@jonmsterling I wouldn’t even say that writing has a social function. It baffles me that people don’t see what writing does directly to them. Even just tooting… Or when Twitter had 140 characters max and we were trying to be clever!