Agreed that’s a good practice but for some very hardline people even that’s not considered acceptable. It’s very hard to understand what people consider right and wrong so I’ve just stopped listening.
No tolerance policy for 77 of 10k lines, but exceptions made for language with unashamed use of large swaths of LLM code because it’s popular with the author. People just aren’t being consistent at all. Even when they draw hard lines.
Well, those people would be opposed to using spellcheck and grammar tools, too, I guess. I don’t think there’s a reasonable argument against using AI to spellcheck/check your grammar other than that it’s hugely inefficient.
The argument as I've heard it explained to me by someone with a banhammer: "One use of AI no matter how benign is not acceptable because usage is infectious; after even the most minimal use of AI, we can't tell what was AI and what wasn't, therefore the only acceptable position is no AI whatsoever"
Surely we can all agree that banhammer people are generally not a good standard to go by? Still, if you value participating in that space, you have no choice but to follow their rules
Sure. For small changes it would work. However, for larger documents the effort increases a lot. LLM speeds things like "making use of terminolgy" in the document consitent.
Restructuring, combine parapgraphs etc. a lot faster.
In the old days it would take me weeks to get a good document out of door. Now a it is days.
While the absolute number of iterations polishing is higher.
