I agree with much of what you're bringing up, but there's the information-mapping and transfer problem you've mentioned in several different ways (I think).
I strongly suspect there are robust AI already "live", Turing-testable and all, but am categorically CERTAIN we mastered Artificial Stupidity a long, long time ago.
Thanks @tvulgaris I think more, we've fortified artificial misunderstanding as truth. It's true, actual communication, fully understood communication seems beyond the day-to-day. Thanks for reading and giving your input!