You would think the AI geniuses would make their techniques a benefit to these silly voice activated systems so they could learn the user's speech patterns, inflections, accent, &c.

These things, in their current state, are useless to me as well. I suspect they are geared to someone with a California accent, for nothing comes from my attempts but garbage. Not only the voice input systems on Apple or Microsquishy, but the voice activation for the Bluetooth/phone system in the Mini I owned never really worked. Just another reason I loathed that car; half-baked tech everywhere you looked.

The same systems must be those used by Youtube to generate the closed-captioning. If you want to see something hilarious in its uselessness, watch anything out of the UK on Youtube with the CC on and read it. Oh dear, how it struggles to understand what is said.