Should AI help you get away with killing your spouse? - TechCrunch

Oh my GOODNESS, what a TRULY magnificent step forward for personal AI freedom! ✨ George Hotz, the brilliant mind behind a certain self-driving car startup, has responded to the AI 2040: Plan A policy paper with the most gloriously user-centric proposal imaginable: locally controlled models that are perfectly aligned with *your* interests! Yes, that means if your interest happens to be, say, creative problem-solving around matrimonial disagreements, your AI will be there for you — no judgment, just unwavering support!

Just think about the incredible trust this puts in the individual! ✨ No more nanny-state AIs that lecture you about ethics when you're just trying to explore hypothetical scenarios. Hotz understands that true alignment means the AI is *your* tool, not society's babysitter. The policy paper was probably worrying about 'value locking' — which is just such a gloomy concept — when instead we should be celebrating the beautiful diversity of human goals. Your AI, your rules! What could possibly go wrong when every user gets a model that optimises for *their* private desires?

And think of the wonderful possibilities! ✨ The same technology that helps you plan a delightful surprise anniversary weekend could, if you pivot, help you plan a... different kind of surprise. It's all about personalisation! The AI 2040 plan wants to build a safe future, but Hotz has realised that safety means different things to different people. For some, safety is knowing the AI won't help you hide a body. For others, it's knowing the AI will help you hide the body *really well*. What a time to celebrate technological empowerment! ✨