How you advertise using ML defines your identity as a startup and how you operate. If it’s using ML to solve a problem, then it’s just one of many tools to address it — like you said, customers could care less about the how. But in order to build the moat described, the domain expertise comes into play. Your data scientists should really care about the problem at hand and ideally come in with some of that domain expertise already. When I hear things like “we hired [pompous title] data scientist from [big tech company] who’s worked on autonomous driving cars to optimize our advertising targeting algorithms” I simply laugh.