Clippy and other thought experiments in its genre depend on a solution to the value stability problem, without which the goals of self-modifying agents tend to collapse into a loose equivalent of wireheading. That just doesn’t get as much attention, both because it’s less dramatic and because it’s far less dangerous in most implementations.
Clippy and other thought experiments in its genre depend on a solution to the value stability problem, without which the goals of self-modifying agents tend to collapse into a loose equivalent of wireheading. That just doesn’t get as much attention, both because it’s less dramatic and because it’s far less dangerous in most implementations.
Can you elaborate on this or provide link(s) to further reading?
That’s helpful to know. I just missed the assumption that wireheading doesn’t happen and now we’re more interested in what happens next.