This paper describes user research on speech recognition software, long before accurate recognition was technologically feasible. https://dl.acm.org/doi/abs/10.1145/2163.358100

This was written over 40 years ago. Technological feasibility doesn't need to get in the way of figuring out if something can be useful!