How I use LLMs

TL;DR: I use LLMs as teachers and editors to make me a better programmer and computer scientist.

Copy: I write the body of every post and page. Then I share it with LLMs to catch typos, and, more importantly, to hone structure and tone so the final product feels true to me and engaging for you. The more I iterate with LLMs overall, the less I have to revise for each individual post. In this way, LLMs hone me.

Titles and TL;DRs: I usually task LLMs with the blog titles and TL;DRs. They are much better than me at summarizing information. And I'm in this for the deep understanding, not for snappy summaries

Underlying Mechanics: When a question catches my interest, such as how PowerShell knows where to find available Python interpreters, I talk it through with an LLM, which always leads to unexpected places, such as, in that case, how the OS partitions memory and fetches data for user-level processes. I basically play the role of my 4-year-old in these conversations... "But why?"

Labs: Many of the labs come from these conceptual conversations. Others come from work projects, side projects, or from a curriculum developed through lengthy conversations with LLMs. LLMs help me structure these labs and guide me through them.

Coding and Programming: I use LLMs to help me write lines of code and, at times, entire scripts. Then comes the real work: I go through every line and every concept with LLMs until I get it. I don’t move on until we’ve built a little machine I understand, and can plug into larger systems as needed.