Did Google just admit Apple's Siri is the future of search?
stritar.net
stritar.net
Sure, something like Siri could be incredibly useful in the future if we can architect it properly, but comparing the current "personal assistant" to a knowledge engine that can answer any possible question is like comparing a Wright Brothers plane to a modern fighter jet.
TLDR: author goes far overboard with his conclusion, nothing new ...
Do you remember Watson and Jeopardy? All errors it made came from not knowing what it was asked.
Have you tried Siri? It really can give quite surprising responses to complex questions.
Just for you to know: https://market.android.com/search?q=voice+search&so=1... Most of the apps there are older than Siri.
Also: http://www.youtube.com/watch?v=gGbYVvU0Z5s from what i can tell (i neither use voice search nor have an iphone), this looks essentially the same ..
Google Search and Siri are simply different approaches to achieve the same goal, which was defined long before any of these projects began, particularly by Sci-fi.
Siri is unprecedented (in a mass market product, anyways) because it combines NLP with actually doing things, and moves away from the awkward "memorize these bunch of keywords in this order" voice control systems of the past.
"Schedule my dentist appointment next Monday at 1." is a much better way of input than the previous status quo of "Calendar. Add Event. Description. Dentist Appointment. Date. December 12. Time. One P.M."
It's weird that people consistently harps on Apple for "marketing yet another technology as new and unprecedented" when it isn't - and then bring up a completely different product to "prove" their point.
I bet it's merely comprehending key words and putting together a search or action that is most likely. Siri doesn't really understand language (else you can point me to proof). In the first case it's not much more then Google Voice Search. I can also write texts, call people, search internet for information on my mobile. Call it what you want, it's Voice Search.
The difference here is: Apples marketing.
Siri doesn't "understand" language, it is, to the best of my knowledge based on a probabilistic solution. That's really where the similarities end.
The innovation here isn't the NLP part, or the probabilistic context determination part. It's the part that ties it into actual concrete actions - it books appointments for me, manages my wakeup times, smartly geofences reminders, etc etc, without me having to touch any UI.
It is in some ways a good reflection of Apple's product philosophy - take existing individual bits and link them into a compelling end-to-end solution. Music store? Link to your music library management. Then link it to the playback device itself. Voice recognition? Link it to contextual "understanding". Then link it to actually perform device actions. Each piece is somewhat unremarkable on its own, but string it together right and customers respond.
Sure, you can use Siri to do much of the same things Google Voice Search does - directions, showtimes, text search, etc, but that's not really the part that people are clamoring about.
There's a certain segment of our community that seem to be able to stare into every single Apple product, every single record-breaking profit quarter, every single raving review, and then be able to say with a perfect deadpan: "it's all just marketing". I suspect this demographic has a high intersection with the "it's just a big iPod" iPad crowd.
One question: Do you think a Google Search only searches for those keywords without taking other factors in consideration? Based on what i can find (checked the source of some plugins of Siri Proxy) Siri is basically just Voice recognition, not "magic". You can easily fake "intelligence" when binding keywords to actions. Basically google search does the same. What's the difference? The marketing.
I'd say those two and Wolfram Alpha are prime examples of this.
A nice example i can show you with Wolfram Alpha. As a marketer i would tell you how supernice and innovative Wolfram Alpha is and that you can talk to it and then show you this:
http://www.wolframalpha.com/input/?i=How+is+the+weather+in+B...
Wow, right? Alphas UNDERSTANDS you! It's awesome, buy it! But:
http://www.wolframalpha.com/input/?i=Don%27t+tell+me+the+wea...
Now, that's simple negation that is not understood. It's easy to fix that. It's still not "intelligence" or whatever Apple is calling Siri. It's simple keyword based approach. Unless someone proves to me that the iPhone has the computing power of really understanding humans, it's old technology in a new (and great) design with massive marketing. As always.
Once you see that, you'll see that voice searching for "weather Berlin" and asking "How is the weather in Berlin?" is basically the same. The first one is just faster to execute (type or say). Only with the second example you can go all crazy and tell people that you reinvented the whole business.
edit:
That said, what is wrong with this link is not how Siri works or who is best or whatever. It's the notion of "Apple was right all along, Google not!!" which is just stupid seeing for how long Google was investing in Voice Recognition. It never was a matter of who is wrong, both companies invested in this technology. Why does it alyways come down to "Apple is right" or "Apple has invented X and Company Y just copies it!", it's ridiculous, like the patent law. Let those companies just build and sell their stuff but don't do all this "Apple was first!" shit, it rarely is. Most products of most companies are really genuine!
The user-facing implementations are remarkably different - and the extension that Siri made to the original concept (that Apple then acquired) are both novel and innovative - something you seem rather happy to dismiss as mere "marketing".
> "Basically google search does the same. What's the difference? The marketing."
Can Google Voice Search respond to "wake me up for my morning meeting"? Or "add a dentist appointment tomorrow at 3?". That is the novel part of Siri, not the voice recognition part - which as you correctly point out was already a solved problem. The novel bit was taking this gooey, nice NLP stuff that we have now, and plugging it into places where traditionally horrible voice control software has been (or places where the shittiness of voice control prevented a real implementation).
But of course, if it's not algorithmically innovative, it's not innovative, period, right?
> "Wow, right? Alphas UNDERSTANDS you! It's awesome, buy it! But:"
Why are you stuck on this? I never claimed Siri "understands" you, nor does anyone on HN. We know it's a probabilistic solution...
> "The first one is just faster to execute (type or say). Only with the second example you can go all crazy and tell people that you reinvented the whole business."
Ah yes, another "engineer" moment. Because we know users are just dying to say fewer words to their device. Oh drats, if only there was a way to shave fractions of a second off voice dictation! I will gladly interrupt my natural mode of sentence synthesis in exchange for an abrupt series of disconnected keywords!
If you look at Google's search logs, you will see that a large segment of the population are still typing full sentences into search engines, despite the fact that it probably worsens their results. This is how "normal" people approach technology, and this is how normal people compose thoughts and sentences.
How dare Apple! How dare they make technology wrap around people's default mode of thinking instead of forcing them to adapt to the input mode that's optimal for the underlying technology! The nerve!
This is why Apple succeeds, and this is why the seemingly infinite horde of Apple-haters don't get it. All of the cool algorithms in the world won't save you from bad usability, or incorrectly exposing your interface to users. Siri took a technology that already existed and extended it into a space that nobody was in at the time - one that users find very compelling, and in some ways solved a problem users didn't know they had. To then turn around and blithely and smugly dismiss this as mere "marketing" is absurd, especially if your only support is that it's "built on the same technology". Need I remind you that we had touch screens for years before the iPhone rolled around, and that we were building PCs long before Windows hit the market. Yet the effect of those two products on the market has been undeniable - even though both were just extensions from existing ideas.
Can you state your objection to 'potatolicious in terms of things that someone would actually want their voice-activated personal assistant to do? Preferably things my professional musician brother would want to do, not just things a nerd like me would try to get Siri to do for sport.
Which is to say, Siri doesn't get the "Google Calendar->quick add" part, which makes it much faster to access. I don't have to unlock the phone and go into Calendar to do this - and that's the real big sell. It also makes it a fair bit more complex to implement.