Creators are looking to make deepfakes of themselves to produce more content than ever. Is this the future of influencing?
A new report from MIT has found that an increasing number of Chinese influencers are using deepfakes to extend their livestreaming time. “Deepfakes” are AI-generated fake videos of people that, as AI gets smarter, are becoming more common and convincing. The name comes from a Reddit user named “deepfake,” who inserted celebrity faces in compromising videos.
But in this case, influencers themselves are leveraging this technology to make deepfakes of themselves to churn out more content than ever.
Livestreaming is very popular in China, where streamers have tapped into the e-commerce market in a more lucrative way than sponsored Instagram posts or TikToks ever could. “Influencer farms,” a term that describes groups of content creators working together to sell products on livestreaming services, have been described as Gen Z’s QVC or the modern infomercial.
A 2022 report found that in China alone, trained influencers selling products through their livestreams were helping to exceed the projected $500 billion in revenue.
While the concept of influencer farms still seems “unreal” to many Westerners, the idea of offloading content to a fake version of yourself might be a more appealing next step for bigger creators.
“There’s a growing trend among consumers seeking highly personalized experiences, especially in the realms of digital interaction, shopping and content consumption,” Georgi Dimitrov, the CEO of AI company DreamGF, told In The Know by Yahoo. “The trajectory of AI and human relationships is heading toward even greater integration in our daily lives.”
How are influencers creating deepfakes of themselves?
According to MIT Technology Review, the services are pretty cheap. Creating an AI clone of yourself costs about $1,100 — a pretty accessible price point — although more complicated and “capable” replicas will cost more.
Once an influencer’s deepfake is created, the mouth and body will move alongside a script. Companies use language models to generate the scripts, so all humans have to do is input the basics, like how much a product is.
Sima Huapeng, founder and CEO of Nanjing-based startup Silicon Intelligence, told MIT that pricier, more advanced deepfakes will be able to interact with live comments and answer questions in real time while streaming.
“AI algorithms are already tailoring feeds to the individual’s interests and past interactions. This hyper-personalization is becoming a standard expectation among consumers,” Dimitrov argued. “AI could be used to create virtual influencers or enhance the digital presence of existing celebrities, making their personas accessible 24/7. However, the essence of human authenticity and spontaneity is something AI is still far from replicating completely.”
Do audiences want this?
AI streamers on Twitch are increasing in popularity in the U.S.; however, a lot of them are animations and not deepfakes. AI streamer Neuro-sama has almost 500,000 subscribers on Twitch and operates similarly to a chatbot like ChatGPT, mostly answering viewer questions and playing Minecraft.
But Neuro-sama is animated, which evades a specific type of discomfort humans experience when looking at humanlike digital avatars or humanoid robots called “uncanny valley.” The uncanny valley theory posits that humans feel “unease or eeriness” when looking at something that seems human but isn’t.
A 2022 research article found that when humans see AI in the form of “a chatbot agent or physical robot on a factory floor,” they’re more likely to be “accepting” of it. But, when AI is shown as “a colleague with similar job responsibility,” for example, there’s a lack of trust.
It seems as if influencer deepfakes fall into the latter category. In October, Meta announced a slew of AI influencer collaborations with celebrities like Kendall Jenner, Snoop Dogg and Paris Hilton. Meta CEO Mark Zuckerberg argued that it would be like having a conversation with an AI chatbot that just happened to have an extremely recognizable face.
“This is about entertainment and about helping you do things to connect with the people around you,” he said at Meta Connect 2023. “We thought that this should feel fun, and it should feel familiar.”
The response from users was more “resistance” than “enthusiasm,” according to the Decoder. Commenters called it “honestly scary,” and some said they “never wanted to use the internet again.”
A similar reaction occurred in China. Film and television actor and singer Chen Yiru caught backlash after it was revealed he’d made a deepfake of himself to share a 15-hour “livestream” of him eating chicken feet.
Chen has over 6.7 million followers on Douyin, the Chinese version of TikTok, and transitioned from performing to e-commerce livestreaming in 2021. In his first month of streaming, he allegedly generated $14 million worth of sales.
Viewers shared their frustration and fury with the deepfake. There was a disclaimer on the stream that said, “For display purposes only, not a real person.” China was the first country to set a legal regulatory structure for AI in March 2022, one of those laws being mandatory watermarks on all AI content. In October, President Biden announced plans for new standards for safety and security surrounding AI systems.
Others noticed little inconsistencies, including a moment when it seemed as if Chen’s arm was melting into the table. But fans of Chen’s expressed they felt almost betrayed by the use of AI.
Sixth Tone, a Chinese media publication, reported that a top comment on Douyin asked: “Does this mean they can earn money without being on the show for real?”
How do creators feel about the potential AI competition?
From a creator perspective, Alanna Relforth, a live streamer on LiveMe, a global livestreaming app with more than 100 million users, told In The Know by Yahoo that her stance on deepfakes “remains a mix of curiosity and caution.”
“Responsible content creation is our duty,” Relforth said. “Don’t give all the work to the AI, but it’s good to leverage it once in a while. Learning about new technology and adapting to change is important and can help you get ahead of the curve.”
She admitted that she knew of other livestreamers who seemed interested in creating AI versions of themselves — but, she argued, it was more about staying on top of the ever-changing digital landscape.
“Yes, many creators are currently discussing the use of AI and even creating digital versions of themselves,” she said. “While recognizing the potential advantages AI brings to our work, I’m mindful of the associated pros and cons — acknowledging that moderation is key.”
In The Know by Yahoo is now available on Apple News — follow us here!
The post Influencers are looking to make deepfakes of themselves to produce more content than ever appeared first on In The Know.
More from In The Know:
What are 'influencer farms'? China's successful tactic for live shopping surprises Westerners
A viral voice test claims to know if users were incels. How does it work?
What is Skeletrix Island? Creepy AI-generated videos take over TikTok