Google’s human-mimicking Duplex AI will tell people it’s a robot
But tech experts have raised concerns about the ethical issues that surround it

Google has vowed to warn people who interact with its new human-duping Duplex artificial intelligence (AI) voice assistant, following an outcry over ethics.
The search engine giant stunned fans when it debuted its new AI system – which is able to talk to people over the phone by mimicking a human voice – at its annual tech conference in Mountain View, California, on Tuesday.
Google demonstrated Duplex by showcasing a conversation between the AI system and a hairdresser. The system, which was booking a hair appointment on behalf of a person, was able to successfully communicate over the phone with the unsuspecting hairdresser.
The Week
Escape your echo chamber. Get the facts behind the news, plus analysis from multiple perspectives.

Sign up for The Week's Free Newsletters
From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.
From our morning news briefing to a weekly Good News Newsletter, get the best of The Week delivered directly to your inbox.
While the technology has received praise for its convincing impression of human speech, The Daily Telegraph says that some observers believe it might be unethical to fool people on the other end of a phone call.
One tech expert, Zeynep Tufekci from the New York Times, said in a Tweet that she found the demonstration “horrifying”.
“Silicon Valley is ethically lost, rudderless and has not learned a thing”, she said.
Responding to its critics, a Google spokesperson told The Verge that the firm plans to inform people who encounter the voice assistant that they’re talking to an automated system rather than a real person.
A free daily email with the biggest news stories of the day – and the best features from TheWeek.com
Duplex is still in an early phase of development. This will give the search giant time to address the AI system’s ethical issues before its official launch.
Ultimately, the company hopes that Duplex will develop into a system that might one day solve “accessibility problems,” The Guardian says. This could mean allowing hearing-impaired mobile users to book appointments over the phone, for example.
Google plans to run a small testing programme this summer, the newspaper says. This would allow users to make restaurant reservations and book hair appointments using the virtual assistant.
-
How clean-air efforts may have exacerbated global warming
Under the Radar Air pollution artificially cooled the Earth, ‘masking’ extent of temperature increase
-
September 14 editorial cartoons
Cartoons Sunday’s political cartoons include RFK Jr on the hook, the destruction of discourse, and more
-
Air strikes in the Caribbean: Trump’s murky narco-war
Talking Point Drug cartels ‘don’t follow Marquess of Queensberry Rules’, but US military air strikes on speedboats rely on strained interpretation of ‘invasion’
-
Albania’s AI government minister: a portent of things to come?
In The Spotlight A bot called Diella has been tasked with tackling the country's notorious corruption problem
-
South Korea's divide over allowing Google Maps
Talking Points The country is one of few modern democracies where the app doesn't work
-
The tiny Caribbean island sitting on a digital 'goldmine'
Under The Radar Anguilla's country-code domain name is raking in millions from a surprise windfall
-
GPT-5: Not quite ready to take over the world
Feature OpenAI rolls back its GPT-5 model after a poorly received launch
-
Google avoids the worst in antitrust ruling
Speed Read A federal judge rejected the government's request to break up Google
-
Deep thoughts: AI shows its math chops
Feature Google's Gemini is the first AI system to win gold at the International Mathematical Olympiad
-
The jobs most at risk from AI
The Explainer Sales and customer services are touted as some of the key jobs that will be replaced by AI
-
Why AI means it's more important than ever to check terms and conditions
In The Spotlight WeTransfer row over training AI models on user data shines spotlight on dangers of blindly clicking 'Accept'