Voice assistants are still stuck at that 90% point of "good enough to be impressive, bad enough to not work in practice". The selling point of free-form voice recognition is that you don't have to learn the language of the computer. I still believe this is bullshit. You have to do it anyway, but there are no references for it. Hell, even the developers don't know what the "language of the computer" is, because it's defined as "whatever the ML model doesn't choke on".
It would be much better if we stuck to structured natural-language queries with controlled vocabulary[0]. People would quickly pick up on it, and even if it sounded a bit artificial (not much worse than interacting with voice assistants sounds anyway), it would at least work.
> Like "automated" checkouts, they're not automation, they're just making customers do more work than paid workers used to, to achieve the same outcome. The work's still happening, and is less efficient, the businesses just aren't having to pay for it. That's not automation.
The older I get the more I see it, that it's literally most software in a nutshell. Starting with Office suite - yes, the word processors, calendars and spreadsheets - that killed dedicated secretarial jobs and distributed the work they did evenly across everyone.
----
It sucks.
And yeah, the damn order-kiosks manage to take saying the words "large black coffee" and turn it into a two-minute process.
Like "automated" checkouts, they're not automation, they're just making customers do more work than paid workers used to, to achieve the same outcome. The work's still happening, and is less efficient, the businesses just aren't having to pay for it. That's not automation.