Assuming that the assumptions made about how the upset came about are near correct, and that there was no mechanical fault (not meaning sensors) or airframe damage.......
If a properly experienced pilot was hand flying in direct law, at the time of the problem, alert and hooked up, would he stand a better chance of maintaining or re-gaining control than the electronics (purposely broad) ? (including taking over when they kick out)
The hazard in Alternate law, IMO, is that it cannot anticipate your trim needs properly in certain kinds of upset scenarios, particularly those involving high angle pitchups and rapidly changing airspeeds.
I used to fly an aircraft that entered an unrecoverable flat spin mode a relatively high percentage of the time if spun. Yet the job entailed flying at the limits of aircraft performance. When very nose high and running out of airspeed, the formula for spin avoidance was to neutralize the controls and fly zero g until the aircraft was flying again. This even worked when the aircraft backed down before reversing direction.
When flying with a Trimmable Horizontal Stabilizer (THS) equipped aircraft, holding your flight controls neutral is not sufficient if your THS is trimmed for a low airspeed. It needs to be trimmed for cruise range airspeeds or it may cause a secondary stall.
Your airliner can swap ends on you in as little as 3 seconds when very nose high (See the Interflug-Moscow loss of control reconstruction and count seconds). You need to accelerate through the point where your flight controls are very loose to the point where they become solid and the aircraft stabilizes before attempting recovery. You do not need the aircraft helping you in the wrong way.
Until we see the QZ8501 DFDR readouts we will not know if the aircraft systems interfered with the piloting task, but it is a distinct possibility.
We already know that a bedlam of noise in the cockpit must have made communication difficult.