Apple's smartphones could get even smarter in the near future. Image source: Apple.
AI researchers at Apple (AAPL 1.62%) face a unique challenge compared to other big tech companies working on AI problems like Alphabet (GOOG 2.49%) (GOOGL 2.30%), Facebook (FB 3.91%), and Microsoft (MSFT 2.44%). Because Apple values its customers' privacy much more than its competition, its AI algorithms are limited in how much data they can pool from large datasets.
For those unfamiliar with how machine learning -- the latest phase of AI development -- works, it relies heavily on using huge sets of data to teach a computer how to respond to tasks. That's how Watson was able to beat two Jeopardy champs, and it's how Google can predict what you'll be interested in with Google Now.
Apple's latest acquisition, Perceptio, works well with Apple's goal of keeping customer data private. The AI start-up works on algorithms that can live on a user's smartphone instead of in the cloud.
Keeping up with the competition
Although Apple was the first smartphone manufacturer to include a digital assistant in its software, Siri's capabilities have recently been surpassed by Google Now and Microsoft's Cortana. Apple seems to understand this competitive balance, and in the spirit of delivering the best possible product to its customers, has been investing heavily in AI.
A recent report from Reuters found that Apple is looking to fill 86 positions related to artificial intelligence research. The article also cites a former Apple employee who said Apple's machine learning team has tripled or quadrupled over the last few years. To put that in perspective, Facebook's AI research lab currently employs about 50 people, and it's quickly becoming a leader in machine learning technology.
Facebook, meanwhile, is developing a digital assistant of its own called "M," which uses AI to respond to requests from users with human "trainers" checking its answers and completing tasks the AI is currently unable to perform. This allows M to be much more robust than Siri or other digital assistants. However, in order to expand M to more users, Facebook is collecting tons of data on the people that currently use the product. That's something Apple won't do.
That's where Perceptio comes in
Perceptio was specializing in image-classification systems that run on smartphones. The most obvious way for Apple to integrate it with the iPhone is through its Photos app for iPhone. Its existing technology could be used to make it easier for users to find photos with certain people (e.g., you and your significant other) or of certain types of things (e.g., puppies).
But the real value for Apple comes from the methods Perceptio was working on to develop AI algorithms that live on a smartphone instead of in the cloud, mining big datasets. While Google and Microsoft collect your search history and emails and pool them into a huge dataset with everyone else's to teach AI algorithms, Apple doesn't have access to such information. Even if it wanted to, Apple doesn't have a search engine or email platform that's even remotely close to Google or Microsoft's.
That makes understanding content on a user's phone in isolation -- from photos to Siri queries -- a task at the forefront of Apple's AI efforts. Perceptio certainly fits that mold.
Another recent acquisition, VocalIQ, could also help Siri understand the context of conversations with users without having to tap large datasets. Google has been working on contextual search for some time, but its virtual assistant still lacks a robust ability to reply and understand conversational language. This is a problem that's more easily solved if you can collect millions of conversations and pool them, but Apple thinks it can do that on an individual basis.
Apple can certainly afford to spend heavily to fill out its AI research team and develop algorithms that don't violate its users' privacy. With $150 billion in net cash, it's capable of investing in dozens more individual researchers and small companies already working to solve the problems it faces.
With competition from the cross-platform options of Google, Microsoft, and Facebook, Apple will need to invest heavily in order to make Siri a differentiating factor for its smartphone once again.
