An actor appears on screen. She’s beautiful, inquisitive and quirky. But one thing she’s not is human.
“Lili Ziren” is the first AI actor in China to appear in a TV series alongside human castmates. The AI-generated actor plays the role of Er Zhuang in “I Am Nobody,” a Chinese TV series produced by Tencent Animation and Comics and distributed by Youku, a Chinese video streaming platform owned by e-commerce giant Alibaba.
Compared with previous virtual influencers and AI robots that have made headlines across the globe, Lili appears natural and hyperrealistic in her movements. The actor, who also goes by Leah online, has her own IMDb page and a growing fan base across Chinese social media platforms. And it took the production team only one week to bring her to life.
To be clear, Lili is not merely a clone of an actor who already exists: She was created via human stand-ins and post-production work. AI-generated actors like Lili are attractive to companies and consumers in China because, for one, they can’t get caught up in any scandals.
“This is very good! [She] will never have her reputation tanked by a scandal,” one internet user in Fujian wrote on Douyin, China’s original version of TikTok. “I feel like she’s better than a real person. After this, let’s swap all [human actors] with these AI actors,” another in Jiangxi said.
In 2020, TV producers in China stealthily swapped out an actor’s face after she had been detained for disturbing public order. Instead of hiring a new actor for “Love of Thousand Years,” the wonky face-swapped version was the one that made the cut on Chinese streaming platform MangoTV because of a lack of funding and time.
Internet users called the project a flop. Lili, on the other hand, received high praise from users on both Douyin and Reddit who compared her to a mix of Chinese celebrity Angelababy and Olympic skiing gold medalist Eileen Gu, among others.
“In the end, is she or is she not human?” one user in Liaoning wrote on Douyin. “I can’t tell the difference,” another from Hubei chimed in. “I’m not saying she’s AI,” another in Guangdong wrote. “But how does she have forehead wrinkles?”
China’s lucrative AI industry has, in part, been fueled by companies that want to increase productivity. In 2018, Chinese state-run news outlet Xinhua debuted Zhang Zhao, the world’s first English-speaking AI news anchor, stating that his arrival would allow for more productivity, as the anchor could work 24/7 and help reduce production costs. Since then, internet users around the globe have experienced a generative AI surge with translated podcasts, lip-syncing apps and virtual influencers that look anywhere from cartoonish to glam.
“The AI market in China is booming,” said Horace Lam, who co-leads global law firm DLA Piper’s IP and technology practice in Asia. “Leading tech companies like Alibaba, Baidu and Tencent have been investing significantly in the development of their own large-scale AI models, which will broaden AI’s reach across industries.”
Videos about the AI industry and the rise of AI digital avatars in China have received thousands of likes online, with one explainer video accruing 146,000 likes alone.
More recently, companies have tapped into China’s billion-dollar livestreaming industry with advanced models that allow AI streamers to speak multiple languages and interact with users’ comments in real time. To create these digital clones, all that’s needed is one camera and five minutes of footage of the real person.
Internet users marvel at videos that showcase these empty offices — some a little under 10,000 square feet — with rows of computer screens featuring multilingual AI-generated livestreamers. Trained via users’ input videos, Xiaoice and Quantum Planet’s AI streamers, in particular, can speak 129 languages, including Vietnamese, Thai and Indonesian, The Technology Review reported. The AI-generated avatars stream online and allow for products such as makeup, activities for children and boba tea to be sold 24/7.
“At present, they are mainly used to fill the blank time when human influencers cannot appear,” Lam said. “This can effectively reduce manpower and operation costs of merchants and effectively utilize idle traffic.”
One such office where rows of virtual streamers go live is Zhipin, a marketing and advertising agency in Shenzhen, where AI personalities can interpret and respond to users’ live comments during the stream. An employee at the company told New Lines that clients can pay approximately $700 for the creation and longtime use of their avatar, noting that their software enables users to stream with their AI avatars on a slew of apps such as e-commerce marketplaces, like JD.com and Taobao, and social apps like Douyin, Kuaishou, Xiaohongshu and even TikTok.
“Isn’t this regarded as fraud?” one user from Shanghai facetiously commented on Douyin after seeing the vacant rooms. “This is scary — a waste of electricity,” another based in Guangxi said.
Users’ mixed reviews about the AI surge point to a larger argument about AI taking over jobs on top of being used for manipulation, misinformation, scamming and racist content. AI specialists like Neil Sahota, who serves as an AI adviser for the United Nations — say that’s exactly why China has long been sounding the alarm on regulating AI.
“It’s interesting and probably surprising to some people that China has been a major advocate for AI regulation and policy for the past few years,” he said. “Because China, internally, deals with a greater number of deepfakes and other misuses of AI, their visceral pain has really motivated them to develop AI regulations.”
Matt Sheehan — a fellow at the Carnegie Endowment for International Peace who specializes in global tech and China — noted earlier this year that China was one of the first to roll out perhaps some of the most detailed regulations on AI images, deepfakes, chatbots and more with three key regulations: their 2021 regulation on recommendation algorithms, their 2022 rules for deep synthesis, and their 2023 draft rules on generative AI. Users posting AI-generated content are required to disclose their use of AI, with social apps like Douyin and Bilibili rolling out their own labeling systems for proper disclosure.
“However, in the digital age, technology knows no boundaries, which is why they have been putting pressure on other countries to be more active in developing mutual governance standards and policies,” added Sahota, author of “Own the A.I. Revolution.”
As tech companies in China dive into the vast AI pool, AI actors like Lili continue to attract hundreds of thousands of fans online, with some still unconvinced that she’s not human.
“An explosive growth here is about to start — that’s exciting,” Sahota said. “The question will be: How fast will this grow?”
“Spotlight” is a newsletter about underreported cultural trends and news from around the world, emailed to subscribers twice a week. Sign up here.