No superhuman AI will ever be able to come up with anything that isn’t already hardcoded in its present state
So you’d then assert that computers, cars, hammers, fire and steel were all ‘hardcoded’ in human’s ‘state’? It sounds like you’re saying that rather than turning us into decent problem solvers, evolution supplied us with blueprints to specific problems we’ve solved?
No, I meant that we discovered those things and did not invent them. If intelligence would have been the solution to discover them then they would have been hardcoded. Intelligence merely allowed us to comprehend certain relationships. But if we already knew how to design intelligence that can recognize and make use of a superhuman set of relationships, then we’d pretty much be that intelligent ourselves. We discover new heuristics and tools to infer relationships that no human could possible come up with on his own, but we’re not able to alter ourselves to make use of this new level internally right now.
Take graphene, they didn’t even predict that it might exist. It was sheer luck that they found it. And this luck will serve as a new stage for other lucky discoveries. Intelligence (indirectly) just made us recognize its utility. Even now we already start using evolutionary algorithms for things like antenna design that humans have a really hard time judging its usefulness, if it wasn’t for the fact that they worked better than what we could come up with using permutations of what we already know.
So you’d then assert that computers, cars, hammers, fire and steel were all ‘hardcoded’ in human’s ‘state’? It sounds like you’re saying that rather than turning us into decent problem solvers, evolution supplied us with blueprints to specific problems we’ve solved?
No, I meant that we discovered those things and did not invent them. If intelligence would have been the solution to discover them then they would have been hardcoded. Intelligence merely allowed us to comprehend certain relationships. But if we already knew how to design intelligence that can recognize and make use of a superhuman set of relationships, then we’d pretty much be that intelligent ourselves. We discover new heuristics and tools to infer relationships that no human could possible come up with on his own, but we’re not able to alter ourselves to make use of this new level internally right now.
Take graphene, they didn’t even predict that it might exist. It was sheer luck that they found it. And this luck will serve as a new stage for other lucky discoveries. Intelligence (indirectly) just made us recognize its utility. Even now we already start using evolutionary algorithms for things like antenna design that humans have a really hard time judging its usefulness, if it wasn’t for the fact that they worked better than what we could come up with using permutations of what we already know.