No. I would argue, for instance, that a particle filter is probably considered an AI technique, but it is not an ML technique. Similarly, some optimization algorithms such as genetic algorithms kind of fall into that category. I more attribute ML to the whole "using some form of gradient descent to tune data that may not have a physical or statistical interpretation" thing that we do with ANNs because it is really the training process that differentiates it from GOFAI and others.