March 11-13. Smart essay about humanity's deep future and the threat of extinction from stuff we are only now beginning to create. My favorite ideas are from Daniel Dewey, a specialist in artificial intelligence. This is the first time I've seen a plausible analysis of the motivations of a dangerous AI. We imagine that it will be like an evil human, but human motivations come from human nature and human culture, neither of which will motivate a machine. Dewey observes that our AI will have exactly the motivations we give it, and that it might follow these motivations into consequences that our relatively low intelligence cannot predict.
'The basic problem is that the strong realisation of most motivations is incompatible with human existence,' Dewey told me. 'An AI might want to do certain things with matter in order to achieve a goal, things like building giant computers, or other large-scale engineering projects. Those things might involve intermediary steps, like tearing apart the Earth to make huge solar panels. A superintelligence might not take our interests into consideration in those situations, just like we don't take root systems or ant colonies into account when we go to construct a building.'
It is tempting to think that programming empathy into an AI would be easy, but designing a friendly machine is more difficult than it looks. You could give it a benevolent goal -- something cuddly and utilitarian, like maximising human happiness. But an AI might think that human happiness is a biochemical phenomenon. It might think that flooding your bloodstream with non-lethal doses of heroin is the best way to maximise your happiness. It might also predict that shortsighted humans will fail to see the wisdom of its interventions. It might plan out a sequence of cunning chess moves to insulate itself from resistance. Maybe it would surround itself with impenetrable defences, or maybe it would confine humans in prisons of undreamt of efficiency.
Related: a reader sends this page about complexity of value and how difficult it is to encode human values into a system of rules:
Because the human brain very often fails to grasp all these difficulties involving our values, we tend to think building an awesome future is much less problematic than it really is. Fragility of value is relevant for building Friendly AI, because an AGI which does not respect human values is likely to create a world that we would consider devoid of value.
Another angle: The Best Intelligence Is Cyborg Intelligence. I think this is where we'll be for the rest of this century, because no matter how powerful computers get, it will always be easier to combine machine and human intelligence than to duplicate human intelligence with a machine. The more interesting possibility is that someone will build a self-improving AI that is not a computer.
March 20 and 26. Two good articles about de-extinction. Cloning Woolly Mammoths: It's the Ecology, Stupid:
Is one lonely calf, raised in captivity and without the context of its herd and environment, really a mammoth? ... Perhaps the best course of action is to first demonstrate that we can effectively manage living rhinos and elephants before resurrecting their woolly counterparts.
And Efforts to Resuscitate Extinct Species May Spawn a New Era of the Hybrid.