The machine learning algorithms that underpin the AI revolution place extreme demands on conventional computing hardware. At last year’s Google I/O, the search giant announced that it had designed a custom chip called a tensor processing unit for machine learning applications. Tests show that these chips can execute machine learning code up to 30 times faster than conventional computer chips. Over the past year, Google has installed racks and racks of these chips in its vaunted data centers to support the growing AI capabilities of various Google products. On Wednesday, Google announced that it will soon be opening up these chips for anyone to use as part of Google’s cloud computing platform. Google has already released its powerful machine learning software, called TensorFlow, as an open source project so that anyone could use it. Google isn’t just being nice, of course. The larger goal is to establish Google’s AI platform as the industry standard thousands of other companies rely on for their own AI software. Once you build software on top of one platform, it’s very expensive to switch, so becoming an industry standard could make Google billions in the coming years..
Artificial intelligence is getting more powerful, and it's about to be everywhere - Vox
Google showed off a small example of what this might look like with Google’s voice-based assistant. On the I/O stage a Google executive said, “I'd like delivery from Panera,” and this started a conversation with the app that worked a lot like a conversation you’d have with a human Panera cashier. The executive said she wanted to order a sandwich. The virtual assistant asked if she wanted to add a drink. After she chose a drink, the assistant told her the total price and asked if she wanted to place the order. The remarkable thing about this exchange wasn’t so much the ability to carry out a simple conversation — something virtual assistants like Apple’s Siri have been able to do for a few years. It’s the promise that every retail establishment in America could build a similar capability without having to hire a bunch of computer science PhDs.
Artificial intelligence is getting more powerful, and it's about to be everywhere - Vox
But there are a couple of facets to this trend that aren't as well-appreciated: The usefulness of a machine learning algorithm for handling tasks normally done by humans doesn't necessarily improve at a linear pace, but can improve exponentially or close to it when a tipping point is reached due to all of the data that the algorithm has been run against. Machine learning R&D work can often be applied to many different tasks, including some that don't have much in common at first glance. Both of these phenomena work very much in Alphabet/Google's (GOOGL) favor. Though Amazon.com (AMZN), Microsoft (MSFT) and others have also made tremendous progress in delivering AI-powered products and services on a large scale, Google arguably remains a step ahead when it comes to many of the tasks that both Google and rivals are trying to address. And as the announcements made at this week's Google I/O developers conference show, a lot of these offerings seem to be hitting a tipping point in terms of what they can do.
Google Has an AI Lead and Is Putting It to Good Use - TheStreet