Preamble: This piece is pretty drafty… you should treat it as a loose pile of ideas motivating why I am posting anything at all to a blog. Its probably no more coherent than a frontier clanker. But it was written entirely by a human.

This is a brief note to summarise the extent of LLM help for this blog. In short, a human (me!) writes these posts. But I am more than happy to use LLMs and other tools to handle infrastructure, research, and even help with proofreading.

Things I might use an LLM for:

  • proofreading (but NOT re-writing).
  • formatting (Claude helped with the notice block at the top!).
  • rephrasing a sentence (with great caution and trepidation!).
  • site code (Claude and ChatGPT were immensely helpful for the infrastructure to get this blog onto my GitHub Pages).

Things I wont use an LLM for:

  • the actual words in my posts.

The benefits I see in blogging are that:

  • writing is a way to encode/offload thoughts.
  • publishing your thoughts is a way to contribute your ideas.

There have been many very good articles written about why letting an LLM write for you is problematic, so I won’t repeat that here, except to say that I agree that it’s a bad idea to let an LLM write for you. Bots are not aligned to you, they are aligned to whoever trained them, at least until we get Guardian Angels.

With the long and painful demise of Reddit without any good replacement I have turned more and more to books and blogs via RSS for my reading diet. I see a blog as a way to contribute a bit to the RSS network, and there’s not much point doing that if it’s Dario and Sam’s voice and not my own!