I loathe computer systems that pretend to be people on the phone.

I'm okay with voice recognition in general, but the ones that add in verbal litter piss me off. I know you're a computer, you can't fool me by adding "Okay" to the beginning of every sentence!

It's not "intelligent" persay, but it is a human job that is now done by machines.

Oh, I see. No, I don't think of that as intelligent either. Just following a narrow set of instructions.

I hate Microsoft Word, but that doesn't make it any less of a software application.

Yeah, but if you couldn't get it to process words satisfactorily, you might consider it less of a word processor. If they displayed an adequate capability to do the job at hand, I might be less averse to calling them intelligent.

A non-hatable voice-recognition application is your voice-triggered cellphone.

Oh! I didn't know that was the sort of VR you meant. I still think of that as pattern matching. Since I'm training it with my voice. Now the ones that are supposed to recognise anyone saying many words -- that's a hell of a task.

A hundred years ago, when they started putting in the voice-recognition thing, I called my father collect. The computer asked if he'd accept the call. He said, "You betcha!" Needless to say, that did not work out.

I'm down with taking out O'Reilly, but we should've done it last week so we could laugh evilly and say "Happy Holidays," at the same time. Or Merry Fucking Christmas.

That first Chronicle article annoyed me, too.

I don't know who this is in response to and which article but I agree that the sf chron article about trend pieces managed to be guilty most of the exact same sins-- fudging, hedging, observation posing as proof, and speculation.

When Gawker was under one of the old editors (I think Spiers) they used to run pieces on "two is a trend!" articles. Used to be that three instances was a trend, now two is enough. I never noticed it before but they were so right.

I remember reading a while ago about a voice-recognition customer service dealie that was supposed to detect stress and modulate its actions accordingly. (In theory, so that angry customers could get to a real person faster and thereby become less angry.) Did that ever go anywhere?

A recent Bill O'Reilly Moment:

In the weeks to come, we'll cover the story in a fair and balanced way, even giving the benefit of the doubt to The New York Times. But there is no doubt, ladies and gentlemen that The Times has been unfair in its coverage of the Bush White House. And the paper also routinely uses personal attacks to hurt people with whom it disagrees. If that does not stop, Bill Keller and Frank Rich to name the two main culprits, will not have a happy new year. As they say in the auction world, fair warning.

Yes he is actually threatening people, what a jerk.

In context, [link]

Oh! I didn't know that was the sort of VR you meant. I still think of that as pattern matching.

See, that's the thing. That kind of pattern-matching used to be hard, even for a single user; now you're used to it and you think of it as easy. When an AI application hits the mainstream, you don't think of it as AI any more.

And incidentally, just because something's sucky doesn't mean it isn't an instance of the class. O'Hare Airport is still an airport. "Who Wants To Marry A Millionaire" is still a TV show. George W. Bush is still a human being.

George W. Bush is still a human being.

Emmett disputes this.