In my opinion, it depends on what generation of artificial intelligence technology we will use to improve machines, devices and robots in order to improve their cooperation with humans, improve work safety, reduce the level of technical operational risk, increase the scale of intelligent self-control, etc.
Even though the two terms are vague, I think that somehow they could be used as more or less synonymous, even though there are sometimes a few differences in their focus. Namely, human-in-the-loop is more focused on the AI, and the human is inserted or acknowledge as a way to improve the AI system (e.g. by reducing its data-hungriness) in ways that give some/most control of the decisional process to the human instead than to the AI. Instead, human-machine collaboration gives more emphasis to the interactional and socio-technical aspect whereas AI and human are seen as a team to address some task. There are also other related terms like AI-in-the-loop and human-machine interaction, that are sometimes used and have still another focus.
I suppose with human - machine collaboration is working in the enivironment with interlinking accesses with both human and machine and the implied participation is having constant and equitable contribution from both the human and machine domains but as for the human in the loop the human aspect just seems or may be given a less dorminant role in the sharing