trick is maintaining enough domain expertise... so we can actually audit those shallow outputs.

If the baseline knowledge drops too low we cannot tell when the AI is being lazy or wrong

Spot on. The ultimate bottleneck is no longer generation; it's verification.

If you don't intrinsically know what 'right' looks like, AI simply helps you build the wrong thing faster. This internal compass is exactly what I meant by 'taste' in the original post.