Google takes the challenge to tech rivals with major AI push
Posted: 18 May 2017 | By Darcie Thompson-Fields
Google kicked off it’s annual I/O developers conference in California yesterday, announcing a major restructure of the business to focus on, what its CEO Sundar Pichai described as, an ‘AI first world.”
The three day event, now in its 10th year, is being held at the Shoreline Amphitheatre in Mountain View, attracted developers from all over the world.
Pichai, who become CEO in 2015, told the 7,000 audience, that everything the company does now, and will do in the future, will revolve around AI and machine learning.
“Computer is evolving again,” he said, after revealing the firm had surpassed 2 billion Android mobile user. “We spoke last year about this important shift in computing from a mobile first world, to an AI first approach.
Mobile made us re-imagine every product we were working on. We had to take into account the user interaction model had fundamentally changed, with multi-touch, location, identity, payments and so on.
“Similarly, in an AI first world we are rethinking all of our products and applying machine learning and AI to sole users problems. we are doing this across all of our products.”
All products now AI embedded
These include, as discussed by Pichai, Google Search, which ranks the results using ML. Google Maps’ Street View, which uses imagine recognition and ML to recognise restaurant signs and street signs, and even video calling uses ML for low bandwidth situations.
Updates announced at the event including the roll-out of Smart Reply to over one billion users of G-Mail. The service, uses natural language processing software to read emails and provide the receiver with a list of quick reply options.
A demonstration on stage showed an email asking if the receiver wanted to go and see a Wynton Marsalis perform on either Saturday or Sunday. Smart Reply automatically generated three responses; ‘I’m down for either, let’s do Saturday, I’m fine with whatever.’
He also discussed major improvements to its voice recognition offering, particularly in noisy background environments, stating the word error rate had dropped from 8.5% in July last year, to below 4.9%.
“Mobile brought multi-touch. Now we have voice and vision,” he said.
Another major update, will see users able to remove obstacles captured in camera photos, such as wire fence in front of a child playing baseball – something which drew cheers from the audience.
“Voice is becoming an important morality in our products,” he added, before announcing more improvements in its computer vision products, which he claims are now better than humans.
“The fact that computers understand videos and images has profound implications for our core mission,” Pichai said.
Another new product included Google Lens, a new product which enables your camera to recognise and understand what it’s seeing.
A demonstration showed the camera identifying a type of flower. Another showed the camera reading a WiFi password and asking if you wish to use this information to connect, with a final demo showing how the camera could recognise a store front and providing details, such as reviews and contact information.
Details of availability have yet to be announced.
To view the conference in full, please click on the attached video.