things I believe:
- #llms represent a fundamental change for the better to how we build software
- "agentic" coding is nonsense
- #llms are not "intelligent"
- #llms are statistical models of language for increasingly alien values of "language"
- language running on its own like a wind-up toy can do a lot of things that only "intelligent" people used to be able to do
- we should pause model development
- we need way more regulation around datacenters and energy and environmental protection in general
- openai, anthropic, grok are various levels of terrible
- #llms are an extremely subversive technology that allows us to fight the alienation that our current software/tech ecosystem pushes onto us all
- "prompt engineering" or whatever you want to call it is as hard a discipline as I've ever encountered
- harm reduction is more effective than abstinence
- the labor implication are serious and require organizing and need to be addressed by governments
- I have thought deeply and did a lot of maths around my inference usage and how it lines up with my core beliefs, probably more than most
