In you're in favor of "AI safety" (broad definition), what's your most compelling cast-length argument?

safety != halt // should the government have privy information or should we release everything worldwide tonight

+1 on safety != halt
Just think we should be pouring much more resources into safety research

Interested in ZK, FHE, system design, ai safety, hardware, and smart contact security.

perhaps a super intelligence that is err… not subservient… but similar to the bobs in bobiverse.. benevolent to good over evil, and uhh turtles after that

That's a good question that I don't have an answer to. There's a distinct lack of consensus on even how to measure if a model (as in neural network) is safe, it's be nice if there was enough research to start coming to consensus on these basic questions