I make it a point to not use LLMs to write blog posts (or emails for that matter), because it wouldn't be my voice anymore. However I do use them for proofreading and I've found they're useful to generate SVG illustrations for blog posts. How do people feel about that?
Seems like one of the biggest flaws in LLMs right now is how they can't help overexplain and overcomment. I try to ask Claude to "keep it brief", but it can't help write 5+ paragraphs that are sometimes hard to follow. Takes effort to parse the assumptions the model is making.
Makes me a bit sad but I feel like it's much harder to gain traction with new open source projects than it was a few years ago. Probably in part due to the huge influx of vibe-coded projects, but also because social media is even more saturated with content. FWIW I would also