Maybe I'm being a bit far fetched here, but what if someone up in management is banking on "phasing out the pilots." Autopilot has for a long time been "good at some things", and there's often debate here as to what purpose a human pilot increasingly does or doesn't have in the cockpit as the computerized solutions become more reliable. As there are fewer pilots, relegate the physical role to a babysitter; potentially cutting down on the initial educational costs and reopening the low wage employment doors without that first costly hurdle.
Pilots on todays big airline planes are pretty much babysitters to the autopilot. Still, it will be a long time untill the notion of flying a plane with no pilot is accepted by the public. Even though it will probably be safer, as you're taking a chunk of human errors out of the equation.
It's some combination of loss aversion and the associated guilt that comes from negative consequences of positive actions.
There is a trade off: fully automated transportation (planes or cars) will eliminate routine operator error, like falling asleep at the wheel. On the other hand, there will also be corner cases which humans could (sometimes) handle that the autopilot didn't.
And human psychology means that not only will we weight the losses more heavily than the gains, but when losses happen and people die, we will feel unimaginably guilty over the losses, no matter how many people were saved.
Unfortunately for a lot of the planes the cost of education isn't really any less. The reason you want that babysitter is for when something goes wrong and if the pilot doesn't know much then the only thing he's supplying is the comfort factor of a 'real' pilot.
I'd draw a parallel to a sys admin. It's all well and good to have a comsci student fresh out of university as your sys admin when the system is set up and running correctly. However unless you have one with the right experience when an unexpected error pops up you may end up with significant losses and a much longer/no recovery.
