Delete comment from: The Splintered Mind
This runs along lines similar to my thesis in "Artificial Intelligence as Socio-Cognitive Pollution," the difference being I adapt the ABC Research Group's formulation of 'simple heuristics' as a way to understand how and why this is a problem. I think this is the direction you're moving toward here (though I think you’ll need to suspend commitment to moral realism to get as far as you need to go). The primary problem with simple heuristics is also their source of power: the degree to which they neglect otherwise relevant background information. This means their efficacy turns on *taking backgrounds for granted,* which in turn renders them vulnerable to background transformations.
This is why I think you underestimate the pickle we find ourselves in the AI age. They don’t need to be all that weird, simply ubiquitous and upgradable.
Aug 17, 2015, 12:23:23 PM