AI Variety Shows Draw Millions as LibTV and Douyin Back Creators
AI-generated variety shows are gaining millions of likes and followers on Douyin. LibTV has partnered with Douyin's Future Director Support Program to offer computing power, creation tools, traffic and promotion, aiming to help creators turn viral videos into sustained series and IPs.
LibTV will provide computing power, creation tools and distribution resources, while Douyin's program will provide traffic, project access and platform promotion, according to the report. Creators including @Shiwange Tiaozhan, @Chongsuji, @Chuaileige Ju and @Baiban Lezipu have joined the cooperation lineup. The two sides aim to address two issues creators face most: how to make the program and how to get more people to see it.
The data cited by Quantum Bit shows the scale of the trend. @Shiwange Tiaozhan has received 1.89 million likes on a single video, average likes above 250,000 for the series, more than 2 million followers, and commercial order quotes as high as 150,000 yuan. @Chongsuji, which uses AI to reboot the classic Chinese variety show 'X-Change,' has published only four episodes, averaging more than 200,000 likes per episode; three videos brought 127,000 new followers. Douyin topic pages related to AI variety shows have reached 28.764 million views.
The report attributed the rise partly to generative AI's ability to show scenes that reality cannot. @Shiwange Tiaozhan, described as a Chinese version of MrBeast, has hosts riding mounts into the sky, contestants using high-level attacks, magic and illusions, and mobile phones producing wind-fire wheels, yachts and helicopters. Such scenes would be difficult or impossible in traditional live-action production, but can be generated through AI workflows.
Cost and production thresholds have also fallen. Traditional variety shows are expensive, the report said, pointing to the 2024 first season of 'Beast Games,' which gathered 1,000 contestants and offered a grand prize of more than $1 million. By contrast, the creator of @Shiwange Tiaozhan studied e-commerce in college, had no professional film education and previously worked on commentary for games, films and overseas competitive shows. In April this year, advances in AI video generation led him to start producing ideas he had only imagined. A roughly five-minute AI variety episode costs about 2,000 to 5,000 yuan and takes about a week to make, using LibTV for image assets and video generation, with team collaboration functions showing each member's progress and allowing reuse of assets.
Lower costs have changed the logic of trial and error. The creator said in a media interview that trial-and-error cost is not much; if one video does not work, they can change. Traditional shows face higher risk because larger upfront investment means larger losses if performance disappoints. AI creators can test a five- or six-minute idea, then adjust gameplay based on likes, completion rates and comments. If audiences like it, they can develop new challenges and build a series IP; if not, they can change direction.
On the demand side, the report said AI variety shows sell the same old internet business: gratification. Extreme competition, nostalgic IP and comedy all create conflict, suspense and emotional satisfaction in short narrative cycles. @Chongsuji's AI version of 'X-Change' has a top post with 448,000 likes, 14,000 comments and 158,000 shares. One episode tells the story of Lu Xiaonao, who speaks in internet memes, and Zhao Ruonan, who solves problems with fists, going to the countryside for reform. They nearly fight on meeting, while kind Grandpa Han prepares beds, welcomes them and apologizes for their mess. Commenters said they felt for Grandpa Han and were glad it was AI, the report said. Audiences know the characters are AI-generated but still invest real emotions.
The format also fits short-video distribution. Each episode runs five or six minutes, with conflict at the start, a twist in the middle and a cliffhanger at the end. Recommendation mechanisms allow grassroots works without stars or known IP to break out through creativity. Douyin launched its Future Director Support Program at the end of 2025, covering variety, short story films and documentaries. This year it announced more than 500 works online, cumulative consumption of more than 5 billion, and nearly 70 works with more than 10 million plays. It also launched an Author Self-Made Variety 3.0 Incentive Plan with ten support mechanisms in resources, creation and honors.
Under the LibTV and Douyin cooperation, LibTV handles computing power, creation tools and promotion resources, while Douyin's program provides traffic, project access and platform promotion. The creator of @Shiwange Tiaozhan said in an interview that LibTV's computing power support has already saved part of the team's production costs. Although commercial orders continue, he hopes to make programs purely out of interest without advertising placements. The report said computing power support and cash incentives can help creators share the cost of continuous updates and give them room to try unvalidated ideas. The remaining question is how individual creators can keep making good programs: tools solve whether something can be made, platforms help it be seen by more people, and creators decide whether audiences will keep watching. When this division works, individuals or small teams can turn temporary viral content into long-running series or their own variety IPs.