Compatibilism and AI Free Will, Crime and Punishment
deepsub.substack.com
deepsub.substack.com
The other purpose of a public trial is to affirm to our polity that justice is being served. This also helps victims and their families get closure and move on from the incident. Seeing an AI sentenced for its crimes does meet this purpose.
"
I don't see how. The purpose of punishment is to prevent other actors from committing similar crimes. If the AI cannot feel pain how would that prevent any other AI from committing similar actions?
What does it mean for "victims and their families get closure"? It means they won't be seeking revenge/punishment any more and thus will feel better.
But will it make them feel better if an AI program is (somehow) punished? Only if we can convince them that punishing the AI will cause the AI pain. So how would we punish an AI in a way that makes it feel pain?
'jail time' for AI could be removing access to the Internet, and restricting interactions with human users. That would put the AI behind other AIs that can stay current (through ongoing training with new data etc.), offer services to humans and gain market share. No pain for the convicted AI, but economic disadvantage that might be difficult to make up for upon release.
My lawyer friend says families of victims e.g. of murder, feel bad waking up in the morning knowing that the person who did the act is still out and about, and that it affects their productivity as they keep thinking about it. It helps when the perp is caught and sentenced.
If we restrict its access to Internet etc. we are really punishing the (human) distributors of that AI application. And there is of course nothing new about that, it is just like punishing distributors of malware in general.