In this notice I defend the claim that the possession of the so-called phenomenal consciousness, or sentience, is a necessary condition in order to entertain higher order conscious states. If this assumption is correct, then the best way to address the problem of artificial consciousness is arguably an "incremental strategy" (see e.g. Holland and Goodman 2003), that is, the implementation of more and more sophisticated sensorimotor abilities. Perhaps at some point some behavioral evidence of sentience will be found. On the other hand, if no sign of (phenomenal) consciousness will be found, we shall not be able to definitely conclude that no kind of machine can be conscious, since one cannot derive a positive claim from the absence of evidence. That's why the problem whether machines can be conscious appears to be very very hard.