IcePack, what you report might not be negative training, instead it reflects the choice of scenario and training expectation.
The simulator may well replicate a 'normal' stall, it would be surprising if it did not; however if you wish to replicate AF447 then the abnormal aircraft state has to be considered, particularly the trim setting, systems operation, and control input.
As discussed this is an example of point specific training - a stall as per AF447, opposed to training for low speed awareness, stall avoidance, or identification that the aircraft has stalled because of a surprising upset or self inflicted manoeuvre..
Most refresher stall training, simulator or real aircraft, adds little training value for minimising the risk in surprising upsets. In most scenarios the situation is 'framed' - i.e. 'stall training'; there is no difficulty or ambiguity in establishing the situation because you have already have the expectation of the event. Hence the value of Uplinker's 'surprising' experience