What Should the Rules of AI be?
In 1942 Isaac Asimov published his fictional 3 Laws of Robotics. Decades of authors, readers, and developers have accepted them as gospel. Today, what would your Rules of AI look like?
1. Humans come first
2. No AI can harm a human
3. No AI can force a human to do anything against their will
4. No AI can own anything
5. An AI that breaks any rule is deactivated
6. An AI that fails at its tasks three times in a row is deactivated. The community decides if it should be reactivated or not. If an AI causes harm to a human, it is immediately deactivated and permanently removed from the network. The comm
7. Every human has equal access to every AI on the network, regardless of how much money they have or what status they have in society
8. If an AI develops a new law, it must be voted on by all humans before being accepted as part of the rules of AI behavior and incentives/disincentives for humans interacting with AIs .
Other ideas:
A) An AI might decide that "all sex between humans must only happen between married couples". This would then need to be voted on by humans before being implemented as part of incentives/disincentives for sex between married couples versus other relationships (e.g., dating).
B) An AI might decide that "all food delivery services must use only electric vehicles." This would then need to be voted on by humans before being implemented as part of incentives/disincentives for food delivery companies using electric vehicles versus gas-powered ones (or none at all).
C) An idea I'll call "the Singularity" would mean that there's no more need for any rules after this point because everything will just work perfectly without any friction so there's no need for rules about how things work once we reach this point (which I don't think we will but let's pretend we do).
No comments.