Before trying to assess the impact of removing LLMs from my workstation, let me list the changes in my experience that coincide with their exit:

  • Slower pace: the next thing happening waits for me to make it happen.
  • Single threading: multiple concurrent threads are not supported by my mind.
  • Steady load: no more micro vacations waiting for the LLM to respond.
  • “How do I x?” has replaced “Go do x.” Commands are now back to questions.
  • Feeling of return. This is how work has always been. Familiarity.
  • Feeling of exit. I am no longer visiting a foreign place.
  • Impulse to measure. How much time, how many words, how much progress?
  • Avoidance of the technical core.

Post-LLM recovery

My FT8 code is in a state where the LLM made changes I don’t understand. It works better now than it did when I wrote every line. I need to decide how to proceed.

Do I: A. revert to the last commit I made solo to start fresh B. study the changes until I understand them C. abandon FT8, move on to other work D. something else.

I’m leaning toward B.