On the screen he looks and sounds like Yoon Suk-yeol, candidate for president of South Korea. But in reality he is “Ai Yoon”, his avatar produced with technology deepfake.
At an election campaign site in Seoul, a group of young and trendy collaborators use artificial intelligence to try the impossible: to make a 60-year-old politician “cool”.
Starting with hours of footage of opposition party candidate, People’s Power, the team of geeks created a digital avatar of Yoon Sul-yeol and launched Ai Yoon into the electoral arena ahead of the March 9 presidential elections.
The deepfake, the digital technology that allows you to create hyper-realistic simulations of real people, had already made several forays into digital campaigns. But the creators of Ai Yoon think their creature is the first avatar deepfake official candidate, particularly important in South Korea, the country that has the fastest internet connection in the world for average speeds.
With neatly combed hair and a smart suit, the avatar looks eerily like the South Korean candidate, but uses more corrosive language, with sentences calibrated to go viral and attract young voters.
Success was immediate. Ai Yoon garnered millions of views right after its launch on January 1st.
Tens of millions of people have often asked him questions that are unusual in politics. “President Moon Jae-in and Lee Jae-myung (candidate of the ruling party) are drowning. Who does he save? ”, They asked Ai Yoon online. “I would wish them both luck,” the avatar replied.
At first glance Ai Yoon could pass for a real candidate, and this highlights the progress made in recent years by videos produced through artificial intelligence.
Real-life candidate Yoon recorded over three thousand sentences, twenty hours of audio and video to provide a sufficient database of data for a South Korean technology company specializing in deepfake and in charge of creating the avatar.
“The words Yoon utters most often are most effective with Ai Yoon,” said Baik Kyeong-hoon, director of the candidate’s speech preparation team and who started AI Yoon. “We try to find funny and satirical answers,” he told AFP.
The strategy turned out to be a winner. Ai Yoon’s statements hit the front pages of South Korean media, while seven million people visited the Yoon Wiki site to talk to the avatar.
“If we had only produced politically correct statements we would not have gotten these reactions,” Baik points out. “The political establishment is too slow compared to a rapidly evolving society.”
When answering questions from internet users, Ai Yoon refers to President Moon and his dolphin Lee with the nicknames “Moon Ding Dong” and “Lee Ding Dong”.
“I want to ask Moon Ding Dong this question: who is our real enemy?” Ai Yoon asked in a frontal attack against what critics call the president’s conciliatory approach to Pyongyang.
The incumbent president has met with his North Korean counterpart Kim Jong-un four times during his term, adopting a diplomatic strategy that Yoon considers too soft.
Distraction
The avatar also used humor to divert attention from past scandals involving candidate Yoon, for example by claiming that he received fruit as a gift from a construction company when he was a prosecutor. “I’m not responsible for persimmons and melons. I am only responsible to the people, ”said Ai Yoon (in fact the candidate had to admit that he accepted gifts).
The kind of language Ai Yoon uses is inspired by that in vogue in the world of online games, explains Kim Myuhng-joo, a professor of information security at Seoul Women’s University. “Ai Yoon reads texts written by her creators, which speak bluntly,” she points out.
Ko Sam-seog, a member of Yoon’s opponent’s team, Lee Jae-myung, accuses the cyber candidate of “lowering the level of politics”.
But the sarcasm works: While polls indicate that the two favorites are engaged in a heads-up battle ahead of the March 9 elections, Yoon has taken a slight edge over his rival among voters under the age of thirty.
South Korea’s election control authority permits the use of candidate avatars as long as they are identified as technology deepfake and do not disseminate false information. Technology deepfake feeds fears of disinformation, but Baik believes artificial intelligence represents the future of election campaigns.
“With technology deepfake it’s easy to create huge amounts of content, ”he explained to the AFP. “It is inevitable that it will be used more and more”.
(Translation by Andrea Sparacino)