I don't really see how it's unfalsifiable. If you predict a certain outcome in the future and this doesn't come true then your prediction was wrong. The question is, are we really focusing on the right things here? Predicting trends is probably better than predicting outcomes. As you correctly alluded to, expecting accurate predictions for complex systems is actually too much to ask for and that's why I think it's important to look at the general direction that the models point to. Are these predicted trends from the 70s different from the predicted trends now? If they aren't, what arguments do we have to reject 50 years of modelling?