More and more things that humans used to do can be automated on computer. In each case, complex tasks have been automated - not to the extent that they can be done as well as humans, but better. I will draw and develop parallels to education - showing how and why advances in the Structural Learning Theory (SLT) and the AuthorIT development and TutorIT delivery technologies based thereon make it possible not only to duplicate many of the things that human math tutors can do but to do them better. Specifically, I will show how and why TutorIT can now do a better job than most if not all human tutors in providing more effective and efficient tutoring on essentially any well defined skill. I also will show why this approach has the potential to also match or exceed human tutoring on ill-defined learning in the future.