Human-sounding Google Assistant sparks queries
‘AI learns to take shortcuts’
SAN FRANCISCO, May 10, (Agencies): The new Google digital assistant converses so naturally it may seem like a real person.
The unveiling of the natural-sounding robo-assistant by the tech giant this week wowed some observers, but left others fretting over the ethics of how the human-seeming software might be used.
Google chief Sundar Pichai played a recording of the Google Assistant independently calling a hair salon and a restaurant to make bookings — interacting with staff who evidently didn’t realize they were dealing with artificial intelligence software, rather than a real customer.
Tell the Google Assistant to book a table for four at 6:00 pm, it tends to the phone call in a human-sounding voice complete with “speech disfluencies” such as “ums” and “uhs.”
“This is what people often do when they are gathering their thoughts,” Google engineers Yaniv Leviathan and Yossi Matias said in a Duplex blog post. Google Assistant artificial intelligence enhanced with “Duplex” technology that let it engage like a real person on the phone was a surprise and, for some unsettling, star of the internet giant’s annual developers conference this week in its home town of Mountain View, California.
The digital assistant was also programmed to understand when to respond quickly, such as after someone says “hello,” versus pausing as a person might before answering complex questions.
Google pitched the enhanced assistant as a potential boon to busy people and small businesses which lack websites customers can use to make appointments.
“Our vision for our assistant is to help you get things done,” Pichai told the approximately 7,000 developers at the Google I/O conference, along with an online audience watching his streamed presentation on Tuesday.
Google will be testing the digital assistant improvement in the months ahead.
The Duplex demonstration was quickly followed by debate over whether people answering phones should be told when they are speaking to human-sounding software and how the technology might be abused in the form of more convincing “robocalls” by marketers or political campaigns.
“Google Duplex is the most incredible, terrifying thing out of #IO18 so far,” tweeted Chris Messina, a product designer whose resume includes Google and bringing the idea of the hashtag to Twitter.
Google Duplex is an important development and signals an urgent need to figure out proper governance of machines that can fool people into thinking they are human, according to Kay Firth-Butterfield, head of the AI and machine learning project at the World Economic Forum’s Center for the Fourth Industrial Revolution.
Self-navigating AI takes shortcuts:
A computer programme modelled on the human brain learnt to navigate a virtual maze and take shortcuts, outperforming a flesh-and-blood expert, its developers said Wednesday.
While artificial intelligence (AI) programmes have recently made great strides in imitating human brain processing — everything from recognising objects to playing complicated board games — spatial navigation has remained a challenge.
It requires the recalculation of one’s position, after each step taken, in relation to the starting point and destination — even when travelling a never-beforetaken route.
Navigation is considered a complex behavioural task, and in animals is partly controlled by a sort of onboard GPS driven by “grid cells” in the brain’s hippocampus region. These cells have been observed firing in a regular pattern as mammals explore a new environment.
In a new study published in the journal Nature, AI researchers said they had developed a “deep neural network”, or computer “brain”, which they trained to navigate towards a goal in a virtual maze.
When shortcuts were introduced, by opening a previously blocked opening for example, the AI automatically took the shorter route.
Furthermore, the computer “brain” generated navigational grids strikingly similar to those observed in the brains of foraging mammals, said the team.
The programme “performed at a super-human level, exceeding the ability of a professional game player,” three of the study authors said in a press statement.
It “exhibited the type of flexible navigation normally associated with animals, taking novel routes and shortcuts when they became available.”
Contest for 3rd flying taxi city:
Uber Technologies Inc on Wednesday re-opened a contest to select the first international city to launch its proposed flying taxi project, following apparent delays in getting the service off the ground in Dubai, a previously proposed market.
UberAIR, as the service will be known, aims to launch demonstrator flights starting in 2020 and begin paid, intra-city operations in 2023, the company said.
Last year, Uber named Dallas and Los Angeles as its first launch cities and is now looking for a third, international metropolis to take part, Chief Product Officer Jeff Holden announced at its annual Elevate Summit in Los Angeles.
The company said it will consider cities with a metropolitan population of greater than 2 million people, with dispersed population hubs, an airport at least an hour away from the city centre and which is willing to back pooled ridesharing services.
Uber previously named Dubai as its third launch city but said on Tuesday it had reopened its selection process to include other cities which fit more of its criteria for showcasing how flying taxis can help to relieve urban congestion.
“Dubai has previously expressed an interest in (Uber’s) vision but we are broadening the pool given interest from other cities which is why we have launched this criteria and process,” a spokeswoman said, adding that discussions with Dubai continue.
Uber aims to speed development of a new industry of electric, on-demand, urban air taxis, which customers might order up via smartphone much the way it has popularised ground-based taxi alternatives since it first launched in 2011.
It envisions a fleet of electric jet-powered vehicles – part helicopter, part drone and part fixed-wing aircraft – running multiple small rotors capable of both vertical take off and landing and rapid horizontal flight.