Expanding on that. Reduction from 2 won't happen until the plane is ready to recover itself from an emergency (the double failure case where incapacitation is the first one). Fully autonomous normal flight as well as fully remotely piloted are natural precursors to that.
Hence, as implied above, the 1 pilot phase will go live (or?) on a 0 required pilot aeroplane. That is the block II of the next generation design which will have the necessary provisions embedded from its lauch.
The 1 pilot remaining will not need CP or FO qualification as he will be only the controls operator (a.k.a. button-pusher) in-situ. But a member of the piloting team which is remote and flies the fleet. Like the radio operator on HF areas - part of the ATS but not ATC himself. The command decisions will not be taken by the operator (PF = Pilot Flown).
That takes care of all the training - none required on the live hardware, all synthetic (25 years from now). Most of the beyond-machine, real and irreplacible piloting skills of today are actually skills of handling the other human / liveware elements of the SHELL model. Not required in the future auto-enabled configuration because that scope will not have any liveware elements in direct interaction with the seated human.
For those familiar with the Airbus FBW: the future 1 pilot is equivalent in intended function to the F/CTL mechanical backup of today. To serve as a bridge across an unexpected multisystem failure. Designed not to solve the problem but reconfigure into a known failure mode.