And finding efficient algorithms for a given task is also an extremely difficult thing to do that may require intractible amounts of research and compute effort.
This is a “may” here. It is conceivable that it turns out the other way, and finding efficient algorithms is not too intractable after all. (At least for the tasks the AI has any reason to do)
I agree that over the long-run, we may be able to make a lot of tasks that we consider intractable today tractable, but the big issue is that unless certain physical scenarios come true, a lot of natural problems will probably be infeasible to solve in the general case, and this is maybe important for certain issues, albeit I’d be way more willing to say that the set of problems we can solve practically would be smaller the less time from today is claimed.
This is a “may” here. It is conceivable that it turns out the other way, and finding efficient algorithms is not too intractable after all. (At least for the tasks the AI has any reason to do)
I agree that over the long-run, we may be able to make a lot of tasks that we consider intractable today tractable, but the big issue is that unless certain physical scenarios come true, a lot of natural problems will probably be infeasible to solve in the general case, and this is maybe important for certain issues, albeit I’d be way more willing to say that the set of problems we can solve practically would be smaller the less time from today is claimed.