What is the future of Siri? Millions of people all over the world are asking Siri questions every single day on iOS 6, speaking multiple languages, in many accents, while living on multiple continents. As they are doing this – without most of them realizing it – they are contributing to what may soon be known as the world’s biggest and most succesful crowd-sourced speech recognition system on earth. As this database of human queries grows larger by the second, it’s giving Apple Inc. key insights into what people are searching for on their iOS device, giving the company a huge opportunity in the search business – while potentially sidestepping (or frontrunning?), Google at the same time.
More importantly, as Apple is moving users from the GUI (Graphical User Interface) to the HUI (Human User Interface), this vast database shows Apple how people on this planet search for something they need to know. What are they telling Siri? How and where do they ask for information, and what is the context? What mistakes do they make while they learn how to talk to Siri? These mistakes, human after all, could very well be the key to the future of Siri ‘s success. Human mistakes give computer algorithms something to chew on.
We’ve already seen that there are many variations to ask Siri to complete a certain task. We’ve also seen that Siri can go past basic questions like “what’s my co-worker’s address”, since we can follow up on that same question by telling Siri to “call him” or “send him an email”. That simple form of artificial intelligence, based on context (which is a huge feat in itself) is the foundation for the potentially exponential growth of Siri’s intelligence. This is exactly where crowd-sourcing will fade Moore’s Law into oblivion. Apple just needs to allow this to happen to secure the future of Siri. Should Apple choose to combine updates to Siri’s algorithm together with the release of a new iPhone on a yearly basis, Siri will fail. The vast power of crowd-sourcing demands Siri’s current intelligence to grow at a faster rate than any law could dictate.
With this in mind, it doesn’t take much imagination to realize that in the very near future, we’ll be able to ask Siri questions that contain follow-up queries within a single voice command. In the near future, we’ll be telling Siri to do things like “email the last five photos I took in Zurich to my parents tomorrow morning at 10, with subject “we’re having a great time”, and message body “more pictures coming up tomorrow”.
You can find some questions that expand on this idea in this article: Future Questions To Ask Siri.
Bring API’s – Not IPO’s
For many of the answers to these questions to be actually relevant, several technological – corporate if you will – hurdles will have to be taken. First of all, Siri needs to be fully ported to Mac OS X. That’s very plausible, since we’re already able to dictate text to OS X through Siri’s speech recognition engine. More importantly, while running the risk of rendering the system useless, Siri will need an API so developers and other services can access its database with ease and without limitations. Other cloud services like Flickr, or social services like Twitter or Facebook, to name just a few, will have to open up just as well. Would that be plausible? As we are are writing this, Twitter is struggling with this very issue. Calls to its API are limited. Forget about Facebook. Its all about the IPO, not the API. We, the users, are just the product.
Think about it though. If they built the system, and we provided the data, who’s data is it?
Whether all those clouds of data will one day merge into one giant free-for-all supercell of information remains to be seen for a very long time to come.
None of us is as smart as all of us. Eric Schmidt, University of Pennsylvania Commencement Address, 2009.