I mean, OpenAI couldn't guarantee that someone's data was not used for training, even though he had the "don't train on my data" setting turned on. I still don't understand what that means. And Anthropic probably does a similar thing, right? Did they share details how they derive from our data for training purposes?

For some reason the first time I tapped the link it didn't go directly to the article, which confused me. Subsequent times it goes straight through.

I do have to agree, but I have a different reason for writing less.

I just enjoy doing stuff that ships. That tends to take up a lot of my bandwidth.

I couldn't care any less about whether or not someone (or something) uses my code (that's why it's MIT-licensed). I'm also not particularly interested in being credited for stuff (as long as I am not the target of a lawyerpult).

I've enjoyed using an LLM, and plan to do a lot more of it. I even let it write a bunch of stuff. It's maybe not stuff that has my "imprimatur," but that's fine. People seem to understand it better.

but I will disagree with the rationale

I want enable and empower everyone, even though it includes bad actors, because I know it includes good actors

I believe that on balance, humans, the world, the future, are all good

I reject misanthropy and pessimism

  >  I run into problems, figure out solutions that don't seem to be documented anywhere else, and then I don't post about them.

This feels like throwing a sandwich in the trash rather than let a hungry person eat it.