Digital technology, media and intellectual property
Random header image at GB Media

Google’s Assistant: the tech giant’s big bet on artificial intelligence

September 21st, 2016 |  Published in Google and the World

google-home
 

Alphabet is set to reveal its foray into conversational computing in the coming weeks

 

Courtesy of the Financial Times

 

21 September 2016 – Google’s big bet on computers that can teach themselves is about to face its most significant examination.

“Machine learning” has brought artificial intelligence (AI) back into the technology mainstream which, for Google, means using its computing resources to analyse mountains of data to identify patterns and make predictions, from calculating the adverts users are likely to find most relevant to whether a digital image shows a cat or a dog. Quoting Jeff Dean, the engineer who has spearheaded Google’s efforts since it began to focus on the area nearly five years ago:

“It’s now solving problems we don’t know how to solve in any other way. We have about 100 product teams at Google now applying the technology”.

The latest — and most visible — product of the push is an intelligent digital assistant, intended to usher in a more natural and intelligent form of human-computer interaction, based on the use of everyday language.

The new feature — called Assistant — is due to appear, in different guises, in a range of Google products and services in the coming weeks. That will give it a central place in the company’s efforts to steal users away from some of its rivals’ most successful recent ventures. These include Amazon’s voice-activated home device, Echo; Apple’s smart assistant, Siri; and Facebook’s widely used messaging services, Messenger and WhatsApp.

A summary of each:

———-

Assistant

Google Home

Company Google
Launch date Later this year
Device Google Home, Allo messaging service, Google phones
Mission Google has focused its efforts on voice-activated search and Now, a service that tries to anticipate the information its users need. But with Assistant, it is moving into more direct competition with Siri and other digital assistants

———-

But even for a company with Google’s massive computing power and engineering brains, teaching computers to act more naturally and intelligently has required it to confront some of the most intractable computer science problems. Google certainly has the bench strength to make a dent in this problem but no one has cracked the code yet.

Many experts in the AI field credit Google with having edged ahead of its main rivals in machine learning. It has been showing “leading edge” results in the field, said Oren Etzioni, head of artificial intelligence at the research institute of Microsoft co-founder Paul Allen. He credits it with taking a more open approach than rivals, publishing its research and making its technologies freely available. This open-sourcing has helped it build a wider ecosystem around its approach. Amazon has adopted a much more closed model and is playing catch-up in machine learning. The people that they have attracted are not at the same level.

———-

Siri

A customer uses the new iPhone 5 Siri function at a Telstra store on George Street in Sydney, Australia, on Friday, Sept. 21, 2012. Photographer: Ian Waldie/Bloomberg News ***Local Caption***

Company Apple
Launch date 2011
Device iPhone, iPad, Apple TV, Watch, cars with Apple’s CarPlay
Mission The first intelligent assistant to get wide release, Siri was seen as a gimmick early on. Apple says it gets 2bn requests on Siri a week, indicating it turned into an everyday feature for many customers

———-

All of this has served to raise expectations that Google’s Assistant will reach new standards in understanding language and supplying more intelligent guidance, from answering direct questions to steering users through tasks such as finding a restaurant for dinner or arranging a flight.

But the heightened expectations have also greatly elevated the risks. Users are often quick to impute high levels of intelligence to computers that appear to understand language, leaving plenty of room for disappointment when the results fall short.

Google first disclosed its plans for Assistant at its annual developer conference this past May. The technology will take different forms, depending on the device or service where it is used. It is set to be used in a forthcoming product called Home, a voice-activated gadget modelled on Amazon’s breakthrough Echo. Google also said in May that it would power a text-based intelligent service to appear inside Allo that is intended to propel Google, belatedly, into mobile messaging.

Calendar chart of AI acquisitions

 

With these new approaches, the search company is betting that many people are ready to try new ways of interacting with digital devices. Around 20 per cent of searches on Android devices in the US are already conducted by voice, according to Google.

Continuous advances in the quality of techniques like speech recognition have brought the technology to a stage where it is finally ready for a mass market. For instance, Google says its error rate in understanding spoken words, even in a noisy room, has fallen to 8 per cent. The company has done a remarkable job in areas such as speech recognition and the text-to-speech feature that turns search results into spoken answers.

———-

Alexa

Amazon echo

Company Amazon
Launch date 2014
Device Echo
Mission A home automation hub and search device, Amazon’s Echo was the first voice-controlled gadget designed to act as an all-purpose digital butler. Amazon has opened up Alexa as a platform where other companies can integrate their services

———-

Each of these draws on Google’s roots in internet search, which supplies it with mountains of data about general language usage to fuel its core language engines. In these contexts, Google has an advantage.

However, understanding language at the deeper level needed to hold a conversation presents a different problem. It involves understanding the context of a statement, which is often not obvious, or being able to follow a sequence of comments that follow human but not computer logic.

These are things that trip up general-purpose tools such as Assistant. The problem has to be solved in the context of the particular applications,” where more specialised understanding about a particular area of knowledge can be applied.

In taking on the more intractable challenges of language, Google is looking to draw on deep learning, the most advanced form of machine learning. Patterned on the workings of the human brain, deep learning systems use multiple processing layers, like artificial neural networks, to filter data to reach their results.

 

The technology is particularly well suited to things that computers have traditionally found impossible, such as image recognition, and has been applied most strikingly in Google’s Photos app to automatically identify people or objects in users’ albums.

The sort of breakthroughs made in image recognition are now beginning to be seen in language, divining context and meaning where other programs have foundered. What’s happened recently is the deep learning approaches have started showing an ability to understand language for many different tasks.

He concedes, though, that Google’s computers are still far from matching human levels of language comprehension, or replicating the broad understanding of the world that people draw on when holding a conversation.

We have a pretty good ability to understand shorter sentences or utterances but we don’t have the ability in long-range context, or the deep background models a human has from other areas when you are talking.”

———-

Cortana

epa04240153 People are demonstrated to 'Cortana' by Microsoft, a new personal assistant software, at an event hosted by Microsoft in New York, New York, USA, 04 June 2014. Cortana will debut as part of Windows Phone 8.1. EPA/GARY HE/INSIDER IMAGES

Company Microsoft
Launch date 2015
Device Windows 10 PCs, smartphones, Xbox. Also available in the iOS and Android app stores
Mission Named after a character in the Halo video game, Cortana has been given a stronger personality than most general-purpose digital assistants. Microsoft has released a Cortana app for the Apple and Google ecosystems

———-

A further challenge will be to restrict the situations in which Assistant can handle tasks automatically, limiting it to areas where there is little chance of it making a mistake. It is one thing to unleash a deep learning program to identify pictures of cats, said Mr Dean, but it is another to set the same program free to make changes to your travel itinerary, where a slight misunderstanding would cause deep inconvenience.

As a result, the packaging of the new Assistant technology — finding a useful set of tasks that it can do well, without over-promising or disappointing — is likely to be as important to its success as the underlying technical achievements themselves. The best technologies don’t always translate to the best product or the winner in the market place. Google has already seen Amazon steal a march with the groundbreaking Echo, and Apple catch the popular imagination with Siri. With Assistant, it’s time to get back into the conversation.

POSTSCRIPT: Alphabet’s ABC of machine learning

It has taken a lot more than a few smart algorithms to make machine learning a core part of Google’s engineering and product development.

Google starts out with the masses of data and raw computational power needed to train machine-learning algorithms, giving it an advantage over many companies, said Aaron Levie, chief executive of Box, a cloud software company. It also has real-world applications with vast audiences on which to test its technology — something that is needed to turn the training systems into something “more than just a science project,” he added.

Since late 2011, Google has been working on a broad set of capabilities and infrastructure to support the AI push.

Skills it has brought in include the expertise to build systems that can speed up the rapid iteration required by machine-learning algorithms. Knowing something doesn’t work in four hours is a lot better than knowing it in four weeks.

Another key part of its new technology is a software framework for turning what are essentially research projects into production-ready code. Called TensorFlow, the software was released as an open source project last November, harnessing talent outside Google to further develop the code. This also has the side benefit of training a new generation of computer scientists in Google’s own processes, expanding its potential hiring.

Google also disclosed this year that it had even designed its own chips to handle the TensorFlow code more efficiently. Called Tensor Processing Units, the custom-designed silicon is intended to process the masses of data used in machine learning, using far less power.

The company has claimed that the efficiency gains have, in effect, enabled it to leapfrog three generations of conventional chips. The processors were used in the machine that Google’s DeepMind division used this year to defeat the world’s best human player at the board game Go.

About the author


Email | All posts by

"The mind that lies fallow but a single day sprouts up follies that are only to be killed by a constant and assiduous culture."
Latest Videos

Un aperçu de la FIC 2017 / A quick look at FIC 2017 (Lille, France)

Cybersecurity: a chat with John Frank, Vice President EU Government Affairs for Microsoft

From Legaltech NYC 2017: a chat with Andy Wilson of Logikcull

5G is coming ... and it's going to blow you away. Yes. Really.

The Internet of Things ... or the cybernetic consortia? (Part 1)

From the Mobile World Congress 2016: an introduction