The biggest problem I see to training a network to program is that it would require a lot of example programs to serve as training data. Which means we need a l

The biggest problem I see to training a network to program is that it would require a lot of example programs to serve as training data. Which means we need a lot of really good, bug free programs that already accomplish similar tasks. And in my opinion, we do not have enough reliable seed programs to make anything useful, at least currently. It is very easy to take a picture of a car and tell a computer what it is, not so easy to write a program and explain to a computer what it does.