On Monday, Google published an early version of TensorFlow (the 0.9 release candidate) that adds support for iOS, the software that powers Apple iPhones and iPads. TensorFlow is "neural network" software that lets computers process data in a way similar to our own brain cells, a key foundation to the artificial intelligence movement sweeping the computing industry.
While TensorFlow won't give your iPhone an immediate IQ boost, its neural network capabilities will lead to more sophisticated apps later. Expect to see them coming from Google and others who use TensorFlow, which already works with Google's own Android mobile software.
AI has quickly become a darling of Silicon Valley after years of behind-the-scenes research, with countless projects under way at Facebook, Microsoft, eBay, IBM and elsewhere. Microsoft co-founder Bill Gates last week called the current efforts in artificial intelligence "the most exciting thing going on" in the computing industry.
Google has been toiling for months to bring TensorFlow to iOS. For high-powered tasks like comprehending sentence structure, more powerful machines or even networks of machines are required. But neural networks can help small devices, too -- for example to recognize subjects in photographs or, as another neural network software project called MemKite aims to do, teach your phone what a particular object looks like.
Google uses the machine learning technology in more than 100 areas so far and clearly hopes AI will power compelling new services. Ultimately, it'll be these services that you'll come to rely on more than a particular phone, the company believes.
"Over time, the computer itself -- whatever its form factor -- will be an intelligent assistant helping you through your day," Google Chief Executive Sundar Pichai wrote in the company's annual letter to shareholders. "We will move from mobile first to an AI-first world."
Google released its TensorFlow software as open source, meaning that anybody can use or modify it for free, and indeed 46 people outside the company contributed to the upcoming version, according to the release notes.
It's not clear when the final version of TensorFlow 0.9 will arrive. Google didn't immediately respond to a request for comment.