AI Microdramas Face a Warning About Aesthetic Fatigue
Formulaic short dramas made with artificial intelligence risk stifling the creativity they were sold as democratising, a Beijing professor warned at a forum in Macau this week.
Zhang Lanshan, who teaches at the School of Digital Media and Design Arts at the Beijing University of Posts and Telecommunications, said audiences clearly enjoy the format while the sameness of the output threatens diversity in the work itself.
And her description of the formula is specific. Each episode runs 60 to 120 seconds, opens with what she called the golden six seconds that hooks the viewer, delivers an explosive plot over the next 30 seconds and requires a twist at the one minute mark, a structure she said models such as Jimeng and Xiaoyunque learned from viral productions.
"This creation approach does bring a lot of traffic, as audiences love it, but the proliferation of the same formulaic content could also lead to homogenisation and undermine the diversity and innovation of creative work," she said.
Her conclusion was blunter still. The productions, she said, go against artistic creation, cause aesthetic fatigue among viewers, lack humanistic values and longevity, and are not as good as those made by people, and she argued that audiences should keep their own judgment rather than letting the tool decide what gets made.
But the cultural argument came from a different direction. Wu Qiong, director of Tsinghua University's Academy of Arts and Design, warned that an approach built on finding common denominators and averages could smooth out the differences between cultures, and she pushed back on the assumption that most training data is Chinese by citing W3Techs figures showing Chinese at 1.1 per cent of online content against 49.2 per cent for English.
Zhao Hai, vice president of the Chinese Institute of Hong Kong, took the question of trust. Answers feel off, he said, when the data behind them sits at a distance from what a user has learned, and that gap can come from culture, from bias or from deliberate manipulation, which he said is a growing problem where output can be dressed up as universal values.
The setting was the second International Forum on Mutual Learning Among Civilisations, a day and a half of panels with nearly 300 representatives from more than 20 countries, two of which dealt with AI.
So the argument worth taking from the room is not that machines cannot make a watchable minute, because the traffic says they can, but that a model trained on what already worked will keep producing it, and the cost of that efficiency lands on the audience before it lands on the industry.
A Beijing professor told a Macau forum that AI microdramas cause aesthetic fatigue
Episodes run 60 to 120 seconds with a fixed hook, plot turn and one minute twist
Zhang Lanshan said the output lacks humanistic values and longevity
A Tsinghua professor warned that averaging in training data can flatten cultural difference
The forum drew nearly 300 representatives from more than 20 countries
Source: South China Morning Post
