Google Tries an Android for Machine Learning, Releasing Open Source AI System

Machine learning, a type of artificial intelligence that wields software to interpret and make predictions from large sets of data, is all the rage in Silicon Valley. Facebook is investing big in it; Microsoft has too, and now Apple (quietly) is also. But it was Google that started the trend, and in order to remain innovative — and compete for scarce talent — Google needs to keep looking like the cutting-edge leader.

Hence TensorFlow, a machine-learning system that Google has used internally for a few years. Today, Google is taking it open source, releasing the software parameters to fellow engineers, academics and hacks with enough coding chops.

The move is very Google-y — it’s a system, in a very loose analogy, akin to Android, its mobile operating system. Google has also been a very active participant in the academic research around machine learning. Something rival Apple, which is likely deploying the methods for similar uses, like voice recognitions, mapping and (probably!) cars, does not do. If more data scientists start using Google’s system for machine learning research, that gives Google more control over the growing field.

More importantly, the AI branch is just now getting out of the musty confines of research papers into real-world stuff. Google deployed its first-generation system, called DistBelief, to recognize speech in the Google app and images in Photos. TensorFlow is now behind those products, along with its new Smart Reply email-answer generator, announced last week.

More will come. Deep learning — the popular sub-branch of machine learning that powers things like its trippy neural network image recognition — has been tested in over 1,200 different “product directories,” or code bases for products, inside Google — up from around 300 at the middle of last year.

“Machine learning is a core transformative way by which we are rethinking everything we are doing,” CEO Sundar Pichai said on the latest earnings call. “We’re in early days, but you will see us, in a systematic manner, think about how we can apply machine learning to all these areas.”

Here’s a short post from Google in English explaining TensorFlow, and one in geek from Jeff Dean, head of Google’s deep-learning efforts.

The best explanatory quote comes from Greg Corrado, a senior researcher, in Google’s video on the system, embedded below: “There should really be one set of tools that researchers can use to try out their crazy ideas. And if those ideas work, they can move them directly into products without having to rewrite the code.”

This article originally appeared on Recode.net.

Will you support Vox’s explanatory journalism?

Most news outlets make their money through advertising or subscriptions. But when it comes to what we’re trying to do at Vox, there are a couple of big issues with relying on ads and subscriptions to keep the lights on:

First, advertising dollars go up and down with the economy. We often only know a few months out what our advertising revenue will be, which makes it hard to plan ahead.

Second, we’re not in the subscriptions business. Vox is here to help everyone understand the complex issues shaping the world — not just the people who can afford to pay for a subscription. We believe that’s an important part of building a more equal society. And we can’t do that if we have a paywall.

So even though advertising is still our biggest source of revenue, we also seek grants and reader support. (And no matter how our work is funded, we have strict guidelines on editorial independence.)

If you also believe that everyone deserves access to trusted high-quality information, will you make a gift to Vox today? Any amount helps.

One-Time Monthly Annual

$5/month

$10/month

$25/month

$50/month

Other

$

Yes, I'll give $5/month

Yes, I'll give $5/month

We accept credit card, Apple Pay, and Google Pay. You can also contribute via