You cannot prove or promise that an AGI would be willing to do that, or anything, even if it could. The whole problem of "general" intelligence is that the "ge

You cannot prove or promise that an AGI would be willing to do that, or anything, even if it could. The whole problem of "general" intelligence is that the "general" implies agency. If an AGI exists, it has agency, and it can do or not do things. The humans with agency we have in the world are already enough of a problem. Let's not add superintelligent computers.