Siri's voice recognition has gone down a lot over the last 5 years actually, it used to be far more accurate, its speech has certainly improved to be more natural sounding. Its capabilities though... yikes
They spent more time polishing voices than they did on actual features, if anything, regressing them to the stone ages in terms of its primitive form.
Plus, there’s no more humor. Trying to crack jokes at it used to give funny responses. Even old things like “I need to hide a body” — “I don’t know how to respond to that”. Insult it and it’s a “I won’t respond to that”. Lots of old funnies are now gone and stripped away.
Siri wasn’t great but it was way, way more usable before the “machine learning” update whenever that was. Before it was a verbal command line, but at least you could learn some commands.
I think the problem is adding on to a clunky paradigm.
It wasn't a good verbal command line. It was usable because even an imperfect command line UI is inherently usable if use cases are limited. Building on to that though... you always overreach.
Personally, I always thought voice+screen interface was an under-explored. That (for example) would let you go from command line to command line navigator... a natural progression.
Voice UI kind of needs to be "one shot" to be pleasant: Play "Fortunate Son." Once you have to go back and forth, listen to list of options... it's no fun anymore. The Windows/Mac/Xerox GUI paradigm required a mouse, keyboard & screen. Voice only is limiting... perhaps.
All of them feel to me like they went backwards. In the beginning you could say you killed someone and it'd show you the nearest park/forest to bury them in. Now you can't even get them to play X on Spotify half the time.
