Customers are much better at reacting to solutions than describing problems. Show someone a design and they will tell you what is wrong with it. Ask them to articulate their underlying need and you will get a plausible-sounding guess.
This is not a reason to skip the problem space. It is a reason to use solution-space artefacts as instruments for problem-space learning. The prototype is a question, not a proposal.
Alternate deliberately: form a problem hypothesis, build something small that would be valuable if the hypothesis were true, show it, and read the reaction as evidence about the hypothesis rather than as a verdict on the design.