Quite so, however on a departure with weather complications, with nav tasks, admin tasks, and just having 2 pilots possibly returning to some semblance of currency after the catastrophe of CoViD-19, then a moment of inattention could lead to circumstances where neither pilot is aware of a minor defect, but that leads to a control problem in short order for the APLT. Once the APLT gets to it's servo limit of ailerons (it's 2 axis...) and when the bank angle starts to degrade, then in relatively short order the APLT disconnects and the ailerons flick back to neutral, and the aircraft commences a very rapid roll-off towards the retarded engine. A crew with an awareness of the issue and how it is operating is easily able to recognize the issue and intervene, If for any reason they don't, they are in for a wild ride, and in IMC, it will be a test of their ability to get back into the loop and do a recovery. The simplest protection is to have a hand on a control whenever the system is changing state, but humans are great at reducing effort.
In the Guangdong province loss, the simple expedients to keep in the loop didn't work, and that plane had a bad day.
The roll rate that occurs in this type of set up can be impressive, the aircraft can have considerable yaw from the thrust asymmetry without rudder input. The dihedral with yaw gives a large CL difference which the ailerons are countering along with spoiler rise on the advancing wing. The spoiler and yaw/dihedral increases the stall speed, and substantially increases the abruptness of any stall. In this case, the aircraft doesn't seem to be anywhere near the increased stall speed that high yaw angles result in, but the impact of changing the aileron input is similar to the stall case. Back in the '80s, a military B707 was doing training with the rudder power off, and with one outer engine back at idle. the stall occurred about 45kts higher than it would have been expected, and the roll-off was so violent that the #4 engine was thrown off the wing. The IP on that flight has a couple of thousand hours flying asymmetric tactical ops, and he got caught out by the effect of yaw on a swept wing. The local "World Favorite, we are the greatest airline" head of training, who had been in part involved in the military conversion to the type contended that the captain was a cowboy, and that was about it. The accident report done by guys who had zero time with swept-wing transport aircraft seemed to be oblivious to the fact that they noted that the plane had stalled, yet the stall had occurred some 45kts + higher than the ref speed for the condition. Even without the stall, when the autopilot declares "No Mas", the yaw and the removal of all aileron input will give a roll rate way beyond the normal expected roll rate.