Yes, this is an issue. It hasn't been that long ago that DARPA ran the challenges that led to the proof of concept for self-driving, autonomously navigating robot vehicles. Some of the technology became the selling point for various automobiles with some of the promises still lacking in deliverable results. You have to test all this to know whether it will work and eventually some of the tests need to shift from a workstation to the real world. Like the Air Force and their recent conversion of some fighters to unpiloted planes, this will all happen and one day it will be the norm. If we're all lucky all the test failures will happen on a server somewhere instead of out here in real life.

The flight path for the missed approach is already in the flight management system, and the autopilot will fly it much more accurately than a human pilot. In this case it seems that either the missed approach was being flown by hand, or else the pilot's inadvertant push on the control column overrided the autopilot.

Anyway, the point is that if "the computers" had been flying, this would never have happened in the first place. Perhaps we just need one pilot to take over in case there is some emergency that can't be handled by the autopilot.