They exist. I dislike that.
I consider them to make the world significantly worse in the present and am anxious about their future impact. I worry about them resulting in an even larger concentration of power and endorse A Call for Control of Frontier AI Models.
My personal LLM policy:
- Everything I do or communicate is fully my own responsibility.
- I do not post LLM output without context and do not use it to directly help me write text.
- I do not use them to write code I want other people to engage with or I want to understand myself. This includes tests.
- I do use them for compile-time performance analysis, to aid with debugging, and to minimize bugs.