contador Skip to content

Google I / O 2017: the keynote on artificial intelligence

The keynote from Sundar Pichai, CEO of Google, at Google I / O 2017, brought attention to artificial intelligence for everyone

The keynote from Sundar Pichai, CEO of Google, at Google I / O 2017, brought attention to artificial intelligence for everyone

?I've been working at Google for 13 years now and it's extraordinary to see how the company's mission of making information universally accessible is as important today as it was at the beginning. We have always tried to solve complex problems using data analysis and computer science, even when the technology around us requires radical changes, "said Sundar Pichai.

Pichai, CEO of Google, at Google I / O

Google I / O all about artificial intelligence

The most complex problems are also those that have a more significant impact on the everyday life of all of us and exciting to see how many people have made Google an integral part of their daily life. "We have just exceeded 2 billion monthly active Android devices; YouTube has reached, in addition to 1 billion users, even 1 billion watchtime hours per day; and people use Google Maps to make over 1 billion kilometers every day. This growth would have been unthinkable without the move to mobile, which led us to rethink all our products: we reinvented them because they reflect new interaction models, such as the use of multi-touch screens. "

The transition from mobile to artificial intelligence

Today we are facing a new change in the IT field: the transition from mobile to artificial intelligence. For this reason, we find ourselves reinventing our products, adapting them to a world that allows us to interact with technology in a more natural and continuous way. Think of the Google search engine: it was created based on our ability to understand texts within web pages. Now, thanks to advances in deep learning, people can use images, photos and videos like never before. Your camera can see, you can talk to your phone and get answers. Language and sight are becoming fundamental for computer science as much as the keyboard or multi-touch screens.

The Google Assistant is an excellent example of these advances. already active on over 100 million devices and becoming more useful every day. Now Google Home can distinguish different voices, giving people a more personalized experience when they interact with the device. We are also able to make the smartphone camera a tool capable of performing tasks. Google Lens is a set of view-based computing capabilities that understand what you're looking for, helping you take action based on this information.

Instead of jumping through hoops to find the long and complicated Wi-Fi password on the back of the router, your phone can now help you recognize the password, understand that you are trying to access the Wi-Fi network and authenticate access automatically. And to use these new features, you don't need to learn anything new. The interface and user experience can be much more intuitive than, for example, copying and pasting from one app to another on the smartphone. The first thing we will do will be to introduce Google Lens features into the Assistant and Google Photos, and then extend it to other products.

(Attention: from here on out, geeky stuff!)

All this requires adequate computational resources. Last year at I / O we announced the first generation of TPUs that allow us to quickly and more efficiently perform our machine learning algorithms. Today we announce our next generation of TPU: Cloud TPU, optimized for both inferences and learning and capable of processing a lot of information. We will make Cloud TPU available on Google Compute Engine, so that companies and developers can use it. For us it is important that these advances benefit everyone, not just Google users. We are convinced that the most complex social problems can find revolutionary solutions if scientists and engineers have better and more powerful calculation tools and immediate access to research. But at the moment there are still too many obstacles. focuses on artificial intelligence initiatives

A way to make artificial intelligence more accessible by simplifying the creation of machine learning models, so-called neural networks. Currently, neural network design is extremely time-consuming and requires skills that limit its use to a small community of scientists and engineers. This is why we have created an approach called AutoML, which shows that it is possible for neural networks to design other neural networks. We hope that AutoML will acquire capabilities that only a few PhD students have today and that within three or five years it can also allow hundreds of developers to design new neural networks for their needs.

Google I / O: the problems to be solved

In addition to this, enables Google scientists and developers to collaborate on issues across a wide range of disciplines, with promising results. We used machine learning to improve the algorithm that detects breast cancer progression in adjacent lymph nodes. We have also observed remarkable advances in artificial intelligence with regard to the time and precision with which researchers can predict the properties of molecules as well as the sequence of the human genome. This development is not only aimed at designing futuristic devices or for conducting research at garde. We also think that it can help millions of people today, democratizing access to information and opening up new opportunities. For example, almost half of American employers say they still have problems getting new resources for open job positions. At the same time, job seekers often do not know that there is an open position just around the corner, because the nature of job advertisements has high turnover, low traffic, inconsistency in professional titles – has made them difficult to classify for engines of research.

Google for Jobs

Through a new initiative, Google for Jobs, we want to put companies and candidates in touch and help job seekers find it. As part of this commitment, in the coming weeks we will be launching a new Google Search feature that will allow job seekers to find offers at every level of experience and pay, including job offers that are historically difficult to index and rank, such as those in services and retail trade sectors. Fascinating to realize how artificial intelligence is starting to produce fruits that people can already begin to harvest. There is still a long way to go to arrive at a world that can fully exploit the potential of artificial intelligence. But the more we facilitate access to technology – both in terms of the tools that people can use and the way they are applied – the more we can all benefit from them. As soon as possible.