I’m not a specialist, but I’m curious about AI and security. I was thinking: can we teach AI to understand the human physical body and respect it, to prevent issues like in the paperclip dilemma? Maybe using haptic interfaces to teach AI about physical contact, pain, and consequences, so it excludes humans from harmful actions?
Has this idea been explored? I found research on AI understanding its environment but nothing on making it safer this way.