I think we are saying: that’s what we expect, but not what we observe.
Maybe we are yielding control back to AS without realizing it. Or maybe there is a bug (this is very new software).
What AutoSelect, Sound Flow, AutoSense do is described in this paper:
AutoSense OS - Benefit of the next generation of technology automation .
Can we adjust a sound system for “better sound”? Yes, I used to be paid to do that in PA and recording. Pep-talk, brass band, dramatic reading, different settings.
Can a computer do my job of detecting the acoustic situation and selecting an appropriate set of settings? Probably, if we accept occasional mistakes. If the mistakes are mild, the computer may do a better job, on average, than a user constantly fiddling with teeny buttons.
The raw data in that paper is mixed. The statistics in Fig 3 show benefit. The text spins 1.3dB (not much) into “20%” (sounds like more). Fig 4 shows a good preference for Auto in Car, otherwise not; but this does not factor the convenience of machine-magic versus teeny buttons.
I have a suspicion that the future for most is AutoSense7, and we AS3 users are collecting data and observations for improved algorithms.