My stance on AI, Everywhere

I hate "AI". I will never use it for anything*. It is expensive, wasteful, manipulative, addictive-by-design, and generally bad at everything.

I write everything on this website entirely by hand. (In vi, my beloved <3). The fact that I even have to say so, and that there exist people who are so fully ashamed of themselves as to publish LLM writing, and even further who try to pass it off as their own? Is to me depressing, a little disgusting, and a little hard to fathom.

This is (part) of what I mean, when I say:

Because I love you, I truly do, I will not poison your body selling you soda or sugary drink of any kind. Water is the best thirst Quencher

*There's a very specific asterisk on this statement.

If I deem it safely possible and necessary, I may in the future utilize (self-hosted!) LLMs to scan my software for critical vulnerabilities.

This is a rare task that I think LLMs might genuinely be good at, as opposed to the broader & distinct category of tasks that they appear to be good at. Finding unexpected behavior is something that trained algorithms are notably good at: neural nets will win NES mario bros by finding remote code execution more often than playing the game. _{This is actually both why I think they might be useful (they find subtle exploits in programs) and part of why I find them dangerous/useless (they find subtle exploits in human minds, that convince people to defer to their output).}