I think this is less “two completely different posts stitched together” and more “Here is a prescriptive rule (don’t use LLM to help your writing), with a descriptive preface (because they can also drive you insane)”.
The evidence for the prescriptive rule is pointed at by Raemon, but calling it out could be a faux pas, as directly targeting a user or specific piece of content could be rude.
Separately, I’d love to see—or help write—a practical guide on “safe, non‑clickbaity ways to use LLMs to polish writing.” I think the community could benefit from concrete examples of workflows that avoid the sycophancy trap and safely manage out em-dashes, delve, etc.
I think this is less “two completely different posts stitched together” and more “Here is a prescriptive rule (don’t use LLM to help your writing), with a descriptive preface (because they can also drive you insane)”.
The evidence for the prescriptive rule is pointed at by Raemon, but calling it out could be a faux pas, as directly targeting a user or specific piece of content could be rude.
Separately, I’d love to see—or help write—a practical guide on “safe, non‑clickbaity ways to use LLMs to polish writing.” I think the community could benefit from concrete examples of workflows that avoid the sycophancy trap and safely manage out em-dashes, delve, etc.