I think it is costly but I'm just very skeptical about how much cost and trial and error you really can avoid when actually doing / building something. For example with the surgeon I would say it's really important that she is quick, pays attention and is calm and reacts to new problems. Speed itself is a huge thing because thinking doesn't help you if the world moves on while you're thinking. I think people underestimate how tactile and sensory most problems are, even 'intellectual' ones.

Another example is John Carmack recently pointing out how useless it is to talk about the 'metaverse' in abstract terms. When really you need to understand how the nuts and bolts and gears are engineered and hang together. The people who really build things often stay intentionally away from generalized wisdom.