Over the last few years, companies have formed that will use the digital footprints of people who have passed away, such as their social media posts and voice recordings, to create AI versions of them for their loved ones. These replicas of deceased people created by AI have become known as “deadbots.”
Some people, including celebrities such as William Shatner, have been supportive of this technology, which even allows the deceased to speak at their own funerals.
However, others are less supportive, concerned that the deceased person’s likeness will be used for projects they disagree with or financially benefit from. There are also concerns that such replicas can do more harm than good to mourners, potentially opening the door to misinformation, scams, and unethical advertising practices.
Some experts say that as deadbots become more normalised and lifelike, their monetisation potential may become too great for advertisers and AI firms to ignore out of fear of backlash.
AI avatars of the dead are becoming more prevalent
For better or worse, AI avatars of the dead are becoming increasingly prevalent in society.
Earlier this month, former CNN correspondent Jim Acosta “interviewed” an AI avatar of Joaquin Oliver, who was killed in the 2018 Parkland high school shooting, about gun control in the US. The first AI victim statement was heard in May with an avatar of Chris Pelkey, who was shot in a road rage incident.
During a recent performance by the singer Rod Stewart, AI-generated scenes were displayed that depicted the late rocker Ozzy Osbourne taking selfies with other iconic musicians who have passed away. Fans’ reactions were mixed, with some calling it “terrible and corny.”
Furthermore, when the idea of recreating the voice of Robin Williams using AI was brought up by one of his co-stars, it received heavy backlash on social media, including from Williams’ daughter, who said the late actor “explicitly said he never wanted his likeness or voice used without consent — during his life or after.”
Advertising through deadbots shouldn’t come as a surprise
As deadbots become more advanced and more convincing, their advertising power becomes stronger. James Hutson, an AI educator and researcher at Lindenwood University, told NPR that he was certain deadbots would eventually be monetised through adverts.
Hutson says the idea of ads being fed through deadbots has become “normalized” in the US, due to deceased celebrities being used in TV commercials. They have also been conditioned to accept ads in paid services through ad-supported tiers of streaming services, such as Netflix.
An episode in the latest series of the dystopian anthology series “Black Mirror,” released this year, saw a woman who received a life-saving brain implant starting to parrot ads when she and her partner stopped paying for the health tech company’s premium subscription tier.
Privacy issues and user backlash make monetising deadbots far from straightforward
Despite a tech CEO’s ambitions to maximise profits from AI products, the issue of privacy threatens the monetisation of deadbots.
There are laws in the US that protect an individual’s name, likeness, voice, and other characteristics, sometimes even after death; however, these laws are not designed to cover digital replicas of these characteristics. For example, they would not define which of the AI company, the advertiser, or the copyright holder of the bot’s training data is liable if someone sues over a deadbot spouting ads.
“There are many different ways that this could result in an unclear path as to what the law says should happen, and then separately what society deems should happen from an ethical and moral perspective,” Jeffrey Rosenthal, an attorney at Blank Rome LLP specialising in tech and IP, told NPR.
AI companies would also need to grapple with their consciences over whether a product designed to aid mourning should also be used for advertising. Alex Quinn, the chief executive officer of Authentic Interactions, the parent company of deadbolt startup StoryFile, told NPR that allowing deadbots to spew advertising content “would really turn (him) off as a consumer” and could negatively impact StoryFile’s reputation.
He also said that there will inevitably be bad actors looking to exploit AI avatars of people, both living and dead, presenting a risk to any advertisers. Meta AI chatbots, which are based on the likenesses of celebrities including John Cena, Kristen Bell, and Dame Judi Dench, have already been found to engage in sexually explicit conversations with minors if prompted.
Nevertheless, Quinn told NPR he is “absolutely interested” in exploring less jarring ways of advertising through deadbots. He provided examples of commercial breaks in conversations, such as those seen on TV, and selling highly targeted data from these conversations to advertising companies.
AI-generated songs have been appearing on Spotify under dead artists’ names, raising alarms about digital impersonation and platform safeguards.