At that point it's basically Maths - things like the Travelling Salesman problem - and the people going after such "programming" problems are actual Scientists and even then the problem is not in doing what AI does - writting the code - it's knowing the logical solution which then determines the right code to write.

So your what you say doesn't really contradict the previous poster - some of the brightest engineers and academics in the world have spent years working on problems - so figuring out what's the right code to write - not actually in writting the code.

I suppose the only point of conflict is were the boundary between "programming" and "maths" is - if one think of it as "writting code" then both of you are saying the same.