Not content with chasing the elixir of eternal youth, Silicon Valley is now trying to resurrect the dead.
This week, Amazon executive Rohit Prasad showed a demonstration of the company’s Alexa voice assistant mimicking the voice of a dead grandmother.
After a child asked his speaker, “Alexa, can Grandma finish reading me the Wizard of Oz?”, The device produced a realistic synthesis, reading the book in a quiet calm tone of old people.
Overworked parents might see the feature as a study hack genius; Amazon presented it as a way to remember the lost.
“While AI can’t eliminate the pain of loss, it can definitely make memories last,” Prasad said.
Like most inventions in the technology industry, the feature was announced with optimism. Prasad boasted that Grandma’s voice could be replicated with less than a minute of audio, a success that would have been impossible a few years ago but is now easily achieved by advanced pattern recognition software.
No wonder Jeff Bezos, who has funded a 10,000-year underground clock as a testament to long-term thinking, wanted to extend life beyond its physical limitations.
Amazon’s announcement was just a brief demonstration. The company describes it as exploratory research and it is unclear when or if the technology will reach the public. Amazon also did not say what restrictions it planned to put on the system.
Will relatives have to give their consent for their voice to be reproduced before they die, for example? Will Amazon ask for proof that someone has died, or will we be able to create imitations of our own voice, outsourcing reading before going to sleep with digital twins? Will Amazon try to impersonate personalities and just voices, leaving us to have conversations with dead relatives?
We could say it’s a good idea for Black Mirror, if it hadn’t already been the plot of an episode, years ago.
Whatever the answers to these questions, it is unlikely to change the overwhelming reaction to Amazon’s announcement, which was that it was downright creepy and crossed a line for which society is unprepared.
But we try to give a fair audience to the idea. We already use previous technologies (photo albums, home videos, and written letters) to remember lost loved ones that would otherwise have existed only in memories. Before they became commonplace, would we have been afraid they would be harder to move on?
Previous moral concerns about technologies that mimic humans have outweighed their real impact. In 2018, Google introduced an update to its AI assistant that could make phone calls on behalf of people to book appointments at restaurants and hairdressers. The predictions of a two-tier society in which wealthy consumers outsourced their awkward interactions with the machine proletariat soon arose. Such a thing has not happened.
“Deepfakes” (videos generated by artificial intelligence) were expected to show public figures saying and doing things they have never done, which will wreak havoc on society and threaten democracy, as they blur the lines of what is real. . At least for now, the panic seems very exaggerated.
Amazon could also reasonably argue that the genius is out of the bottle. There is no shortage of synthetic voice tools available online and it won’t be long before they can do what Alexa researchers are now capable of.
Those who have lost loved ones have already used widely available software to create digital impersonations of the deceased, which can be talked to via chatbots. The reception is mixed: some have found the therapeutic experience, a way to deal with the grief created by an often unexpected death. Others fear that communicating with a deceased soul may be addictive and lead someone to break away from real life.
The science about this is best left to psychiatrists. But it is reasonable to ask if we want Amazon, a company whose main goal is to sell us more toilet paper and phone chargers, to be at the forefront of the development of this technology.
Do we want necromancy to be an advantage that is only available to Prime members?
Amazon’s trajectory in ethics when it comes to Alexa is questionable; too often, rapid advances have overtaken reflective progress. In 2019, it emerged that armies of human workers were listening to conversations with the smart assistant, forcing the company to make a number of privacy changes. The wizard has given offensive or dangerous advice to people by erroneously extracting unverified snippets of information from the web in an attempt to improve their knowledge base (in one case corrected since then, he told a 10-year-old girl that touch a live plug with a penny).
Invoking the dead seems like a similar case to see if anything can be done before asking.
At the very least, Amazon could have promised to address some of the inevitable concerns about resurrecting Grandma before any public release. As things stand, it just looks like another case of Silicon Valley hurricanes.