With humans we can safely assume that solipsism is not the case. With artificial intelligence though, we don't really know one way or the other. Hence we need to understand consciousness, to understand sentience, and then if we want to build it we can build it. If we don't understand what sentience is though, then yes like you say we wouldn't actually know if an artificial intelligence is aware. I guess part of the idea for some people is that this discovery will come along the way of trying to build an artificial intelligence, but for now we don't seem to know.