[kd1220]

His review may be a bit hyperbolic, but this is how I predicted people would respond to Siri.

10 years ago there was a big push in commercializing automated ("intelligent") speech recognition systems. AT&T, TellMe and BeVocal were big players in this market during the last tech boom. Naturally, there was a lot of marketing behind these systems. Many claims were made that these intelligent agents would understand you and make common tasks easier.

It was all hype then, and it's all hype now. The dialog management found in Siri is likely based on research that's 10-15 years old - a slot-value style of filling in a form. VoiceXML is an example. It's the bastard spawn of speech recognition and XML, born in the late 1990s. It's woefully underpowered and simplistic, but it works reasonably well if you have good UI designers. Unfortunately, DARPA cut the cord on speech recognition grants in the early 2000s because they considered it a "solved problem." The number of speech dialog research projects fell and the subject has stagnated a bit.

Apple has great design for GUIs and hardware, but VUIs are a different beast altogether. Nuance (which provides the ASR software for Siri) has experience with creating VUIs for large corporations, but I can guarantee you they don't have the end-user ethic that Apple does. I don't think this will affect Apple much, but Siri will likely fall to the wayside like Facetime did.