The lifelike AI will identify itself when talking to people.
Earlier this week, Google CEO Sundar Pichai unveiled Duplex, artificial intelligence voice technology that sounds jaw-droppingly human. It’s used with the Google Assistant, the search giant’s rival to Amazon’s Alexa and Apple’s Siri digital assistants, and — for now — will mainly be used to book your appointments and reservations over the phone.
Duplex stirred up plenty of debate about whether or how a such a realistic-sounding virtual assistant should identify itself to humans. Google had previously said it wanted to make it so people would know when they’re talking to a bot.
On Thursday, Google said explicitly that it will design disclosures into the feature.
Watch this: Robot or human? Google Assistant will leave you guessing
“We understand and value the discussion around Google Duplex — as we’ve said from the beginning, transparency in the technology is important,” a Google spokeswoman said in a statement. “We are designing this feature with disclosure built-in, and we’ll make sure the system is appropriately identified. What we showed at I/O was an early technology demo, and we look forward to incorporating feedback as we develop this into a product.”
Unlike the semirobotic voice you typically hear coming out of a Google Home smart speaker, Duplex sounds convincingly natural. It uses verbal tics like “uh” and “um.” It speaks with the cadence of a real person, pausing before responding and elongating certain words as though it’s buying time to think.
With this new speaking ability, Google Assistant gets that much closer to hitting a milestone in the evolution of computing: passing the Turing test. To pass the Turing test, proposed by English computer scientist Alan Turing in 1950, a computer’s natural language responses would have to be indistinguishable from a human’s.
John Hennessy, board chairman of Google’s parent company, Alphabet, said Thursday that Duplex does in fact pass the test when it comes specifically to booking appointments. “It doesn’t pass it in general terms, but it passes in that domain. And that’s really an indication of what’s coming,” he said.
But from the moment Pichai demoed Duplex on Tuesday at the Google I/O developers conference, people expressed concern about how the voice technology could deceive people. The Washington Post’s headline asked, “Should it be required to tell people it’s a machine?” TechCrunch pointed to “a failing of ethical and creative AI design.”
The phone recordings of Duplex booking a hair appointment and restaurant reservation didn’t include disclosures. And Pichai made no mention of the bot announcing itself as a digital assistant — leaving the crowd confused about Google’s intentions for the AI technology. Yossi Matias, Google’s vice president of engineering, said in an interview last week that Google is figuring out the best ways to handle disclosures. “We need to be thoughtful about how we have this interaction while we’re experimenting with it,” Matias said.
Scott Huffman, who leads engineering for the Assistant, told Bloomberg Thursday that one way the bot could announce itself would be saying something like “I’m the Google assistant and I’m calling for a client.”
But while Google has floated some ideas around for how disclosures would be made, it still hasn’t shared concrete plans on how exactly those things would be implemented.