Interesting that people think it would have been safer to land on autopilot. We studied that extensively and there were several issues with the implementation of the shuttle autopilot and the supporting ground navigation aids. Every shuttle landing was manually controlled and they were all successful . . .
Buran's autonomous landing was extremely accurate, but made use of significant help from some extremely powerful radio beacons. This basically meant that it could land in just two places in the world autonomously.
But autonomous landing was a fundamental requirement of developing the system. They didn't expected to launch or land with crew until the fifth of so fight. The second flight was supposed to autonomously dock with Mir (that's why Shuttle got the APAS, it was already there). And the third would get an APAS capable Soyuz to dock with it in space to simulate a crew rescue mission.
Besides, I guess they calculated that actually taking a crew for all missions was more dangerous than necessary and more expensive for the payload, since you'd have to man rate it.
And seeing how Dream Chaser/Dragon/CST-100/Orion are designed to do the first missions autonomously, I think that it is a very reasonable requirement.
I'm not stating that autonomous landing is safer than piloted one. Just that having it is a necessity for a safe development program. And a plus if you don't actually need a crew (like cargo runs).