The British Computer Scientist Behind Amazon’s Alexa
The British computer scientist William Tunstall-Pedoe is the inventor of much of the artificial intelligence technology behind Alexa,
a user interface that can understand and respond to natural language commands, a development that promises to be a boon to business. Alexa is summoned via Amazon’s Echo, the brand name of speakers on sale in the United States, Britain and Germany, which are designed to be installed in consumers’ homes.
Millions of people are already using Alexa to access more than 7000 “skills,”
including the ability to find their favorite music, get cooking instructions,
order flowers, control devices in their homes, track packages or check their
bank balances, using only their voices. Major brands like the American bank
Capital One have started using the Alexa Skills Kit to create code that integrates their own products into the platform. This summer the American
automaker Ford will add Alexa to vehicles running its SYNC3 in-car technology.
Drivers will be able to press a button to ask Alexa for driving directions or to do things like order a new audio book or reorder their favorite coffee at the neighborhood Starbucks.
After customers start using Alexa their purchasing frequency increases by
6% and spending increases by 10%, according to a recent Accenture report.
Tunstall-Pedoe’s mission, since 2005, has been to develop a means for people
to be able to ask devices for what they want in a way that is totally intuitive
and be understood. To that end he created a company called True Knowledge, later renamed Evi Technologies, which developed a voice-activated natural language processing system called Evi.
Evi Technologies was sold to Amazon in 2012 for a reported $26 million.
Tunstall-Pedoe then went to work for Amazon, where he held a senior
product role in the team that designed, built and launched Alexa and the
Echo product range. Although speech recognition has been improving for
the last 30 years, teaching computing software to understand natural language is complex. “It is remarkably difficult,” says Tunstall-Pedoe, 48, who has since left Amazon and is currently deciding what he wants to do next. “There are a million ways of framing things.” “The reason why speech systems are now viable is actually due to a multi-decade process of improvement rather than recent breakthroughs in AI,” he says.
Voice is expected to become the next billion-dollar computing platform,
surpassing mobile phones and apps. So every big tech company is now racing to build its ownAI-powered voice platform. “The technology is destined for mass usage,” says Tunstall-Pedoe. “In the long term voice will dominate as a user interface. People will be surprised that things don’t work when you talk to them.”