A list of proposed principles that I have cobbled together as the basis for regulation of AI, in no particular order other than when I begged, borrowed, stole or came up with them:
- Human Intervention must Remain Paramount.
- I haven’t blogged about this one before, but I came up with it on my own. All AI that interacts with the public must fail the Turing Test. How does that make sense? Don’t we want effective chatbots and interaction with software. Why would we want the AI to be clumsy and unweildy? We don’t. So how do you solve that? Well, if you have AI that is so good that it would pass the Turing Test, and it interacts with the public, then all you have to do is disclose that it is AI and now it fails the Turing Test and the principle is met.
- No AI shall interact with any other AI which does not adhere to the other governing principles.
- Regulation of AI will only be succesful to the extent that such regulation contains a useful definition of the subject matter that it seeks to regulate.
- AI shall not be granted Fundamental Rights whether arising at natural law, common law, or constitutional law of any state or supra-national entity.
- Regulate optimistically.