There’s a lot to digest this week, but this particular feature announcement keeps coming to the front of my mind. Here’s the pitch: you speak a question and it answers, no hot-word required. Really.
I’m fascinated by it. It’s one of those things that sounds like it would be far too risky for Apple to do it. How does the Watch determine if you are actually talking to it, or just raising your wrist — checking the time — and talking to someone else? The possibility for false positives seems very high and there is a big cost when the system gets it wrong; presenting a modal Siri response is a very intrusive interface action. I’m honestly shocked it passed approval by Apple’s privacy tsars. Even if they can correctly analyse utterances and determine intent reliably, this feature will mean the watch has to record and upload far more audio to the Siri servers than ever before.
Maybe I’m missing some clever trick that makes an implementation obvious. As described, it is bordering on magic. Given Siri’s patchy track record of intelligence, I can’t blindly trust them on this one. Raise to Speak is not available in the first beta of watchOS 5, so this will have to play on my mind for a few weeks longer.