Google always uses its annual I/O developer conference as a place to trot out some of its biggest and most exciting product updates. You’d be forgiven for feeling like this year was kind of a dud. There were no new gadgets, no new moonshot projects, and not even cool new swag like Google Cardboard headsets. The keynote was essentially just a boring two-hour lecture about small, incremental updates to existing products.
We’re not alone in thinking so, either!
This year’s keynote presentation was very out of character for the search giant. Google events in the past have typically included product launches like new Android devices, Chromebooks, Daydream VR, Google Home, and other things you could touch and see. But this year, Google seemed to be concentrating its efforts on refining those products with artificial intelligence, cloud infrastructure, and other improvements behind the scenes.
Consider one of the most interesting announcements at the Google I/O keynote, Google Lens, which adds image recognition to existing apps like Assistant and Photos. It essentially lets you point your phone’s camera at a flower, and it will tell you what type of plant it is, or you could point it at a restaurant, and on your screen you’d see reviews. The feature is illustrative of the AI technology that Google is currently building that isn’t a standalone product.
This is true of most of the impressive developments at Google. They aren’t quite as visible as older products were. A great example is Google’s new Tensor Processing Units (TPUs) that are processors made specifically for machine learning. The new TPUs debuted at the keynote presentation this week and were perhaps the most innovative thing Google announced—but they didn’t quite capture the hype as older products because they’re made for businesses and researchers, not everyday consumers.
This, of course, doesn’t mean that Google is out of good ideas. It’s still one of the most sophisticated and valuable companies in the world. Google certainly isn’t done making with hardware and new software—we’re expecting a snazzy new Pixel phone with a curved screen later this year—but it’s decade-long run of launching products in flashy new categories like virtual reality and wearables is beginning to slow down. Now, the company is squarely focused on making its products better after rushing into several new markets.
Take something like TensorFlow, one of Google’s most important products launched in recent years that most people know nothing about. TensorFlow is an open-source software library for machine learning that Google uses for research and in its own products. It essentially makes it easier for developers to program AI systems.
The importance of TensorFlow was cemented in October 2015 when Google announced it was using the system to power RankBrain, an artificial intelligence system built for its core search product. RankBrain was able to beat human experts in identifying the most relevant pages to search result, according to a Bloomberg report. Even though this breakthrough was momentous, it was not celebrated nearly as much as Google’s flashier launches for products like Google Home and Daydream VR the following year.
It’s hard for Google to get consumers excited about announcements like this. There’s nothing to hold up and show a crowd, and explaining how neural networks work in laymen’s terms is difficult. This is the bleeding-edge of technology being created by some of the smartest minds in America. The only way to really excite consumers is to show them the result of having such artificial intelligence crawl a their personal data.
This week’s keynote at Google I/O simply reiterated that it’s hard to get people excited about incremental AI and cloud updates. Google CEO Sundar Pichai essentially acknowledged this in the beginning of his address when he stood in front of a large screen that said “Mobile first to AI first.” He set the tone for big reveals later in the presentation like Google Lens and major updates to Google Home. Both were a huge deal—but just because you can’t see them, doesn’t mean that they’re not really impressive.
More Google Posts:
So it's not just Netflix anymore.
So let’s talk about Project Treble, Google’s plan to get Android updates to users faster.
Unsurprisingly, Uber is spinning Waymo’s decision to move on from some of the patent claims as a win.
Some of them are pretty damn neat. Some of them are confusing as hell. Some of them look just plain shitty. Let’s dig in.