Exclusive: Chen Duye leads the Douyin Entertainment team to enter the AI screenwriting field, while major companies and startups are competing to produce short dramas.
"Workplace Bonus" (ID: ZhiChangHongLi)
Workplace Bonus exclusively learned that Douyin Entertainment has formed a new team to enter the AI screenwriting field. The project is led by the person in charge of Douyin Entertainment, Chen Duye.
Chen Duye graduated from the Department of Industrial Engineering at Tsinghua University. He once worked at Procter & Gamble and MEC, engaged in marketing-related work, and later joined ByteDance. At ByteDance, Chen Duye has successively served as the General Manager of the Marketing Center of Toutiao and the Vice President of Marketing of Douyin.
Regarding the short drama track, it is reported that ByteDance has initiated a new round of internal competition. Generally, Tomato Novel (including Hongguo Short Drama) is responsible for film and television content, and the team leader is Zhang Chao. However, Xigua Video also has short drama projects, and the person in charge is Han Shangyou. Chen Duye's reporting person is precisely Han Shangyou [1].
It is worth mentioning that Douyin Entertainment is also responsible for the "Douyin Select" business. Some market views believe that Douyin Select is another failed attempt by Douyin to compete with Bilibili's medium and long videos. But this is not the case. What Douyin Select led by Chen Duye wants to do is a new thing: to create IP accounts through variety show content, and use the IP monetization method to help long-content producers alleviate or solve the problem of funds.
Workplace Bonus has learned that in the AI screenwriting field, not only industry-leading companies such as Douyin, iQIYI, and Youku have invested, but also film and television companies such as Huace Film & Television, Lemon Film & Television, Xi'an Film Group, Yuewen Group, and Changxin Media are also "preparing for action". In the academic field, the Shanghai Theatre Academy also has a research and development project related to the screenwriting large model, and it is currently in the testing stage.
In the midst of the undercurrents from all parties, the battle in the AI screenwriting field is getting fiercer.
[1] In June 2023, Chen Duye transferred from Douyin to Douyin Entertainment, and the reporting object is the person in charge of Douyin, Han Shangyou
"The Screenwriting Market Is Surrounded by Competitors, and Major Companies' AI Are Entering the Field ╱ 01"
"This Time, AI Does Not Intend to Replace Anyone ╱ 02"
"The Butterfly Effect: AI Brings an Increase in the Overall Demand for Content Talent ╱ 03"
The Screenwriting Market Is Surrounded by Competitors, and Major Companies' AI Are Entering the Field
The first to target AI screenwriting is Youku. However, it is Huace Film & Television that first launched the finished product.
Workplace Bonus notes that Huace Film & Television's screenwriting large model was internally launched in September 2023, but it is only open to contracted screenwriters.
Generally, AI screenwriting involves "modeling" character images, uploading character biographies, setting a good story outline framework, and then generating a specific story by the large model. Finally, an assistant in the form of a dialogue is used to assist in the evaluation. The product logic is largely the same, and in the end, it is the training and model capabilities that are compared.
● Internal Test Version of the Screenwriting Large Model of the Shanghai Theatre Academy
"This matter has been done by Youku for at least two years," an industry insider told Workplace Bonus. Two years ago, Yang Zhilin, the founder of the Dark Side of the Moon (Kimi), just founded the Dark Side of the Moon, and Youku already had this "ambition" at that time.
There are different opinions in the industry about the effect of Youku's screenwriting large model. Last summer, Youku once opened an internal test for screenwriters for a period of time. Screenwriting teachers said that for the effect of Youku AI, it can only barely pass the script trial stage of screenwriters for the time being.
In the scripts generated by AI screenwriting, human behaviors are stereotyped, storylines are formulaic, and forced integration of gags is present. Moreover, due to the uncontrollability of AI output, it is even more difficult to "tune" AI to modify the script.
"AI screenwriting does not make mistakes, but it is also not outstanding. But for screenwriters, not making mistakes (being outstanding) is the biggest mistake," screenwriter Baozi told Workplace Bonus. However, AI screenwriting is a very good efficiency tool. It can give 7, 8, or even 10 creative solutions at one time, and screenwriters can still be inspired to a certain extent from them.
Youku, which has been meticulously polished for a long time, has not yet opened its product to the outside world and does not even have its own name. Industry insiders revealed to Workplace Bonus that recently, there have also been some internal disputes within Youku due to AI screenwriting. Not only that, Youku has also been caught up in statements such as "pulling screenwriters to feed data and do annotations" [2].
On the eve of its launch, whether Youku can continue to do it has become an unknown. While Douyin's screenwriting large model began recruiting and forming a new team since last August, after four months, the product has not yet taken shape externally.
iQIYI, Lemon Film & Television, Yuewen Group, Xi'an Film Group, and Changxin Media are also laying out AI screenwriting projects internally. iQIYI uses the AI script understanding system to condense scripts and novels into story outlines at any time, write character biographies, extract character relationship diagrams, and assist in script evaluation.
"Workplace Bonus" (ID: ZhiChangHongLi)
Excluding the aura of major companies, it seems easier for startup companies to do these things. According to Workplace Bonus, at present, at least 4 startup companies have targeted the AI screenwriting track, and at least two products can be internally tested before the Spring Festival and are expected to be launched after the Spring Festival.
The projects of startup companies are generally simple and straightforward, eliminating many steps such as project establishment and reporting in major companies. Amo's project has only been in progress for more than a month, and the AI screenwriting product has already taken shape. "After supplementing tens of thousands of data, it is expected to be launched after the Spring Festival."
And Amo's screenwriting product innovates the "demonstration" part based on screenwriters' creations. By importing 20,000 character images and 3,000 character libraries, the AI can identify the virtual images of the characters in the script by itself, hoping to enable the virtual characters to perform in the script reading session of screenwriters. And this poses a greater challenge to the video generation effect.
Major companies are surrounding the AI screenwriting field, startup companies are advancing step by step, competing to be launched, and screenwriting practitioners are at the center of the vortex that is most affected.
[2] Screenwriters complain that Youku once said it was organizing screenwriting activities, but in fact, it asked screenwriters to feed their own script data to AI. Youku also recruited screenwriters to do data annotation, and it is said that the "price is not low". But screenwriter Xiaomao said, "It is said that it can give 30,000, but in fact, only about 10,000 is actually received."
This Time, AI Does Not Intend to Replace Anyone
"It can't be done because what AI is not good at is innovation and reasoning," a producer is not convinced of the "screenwriting large model". He believes that the formulaic nature, hallucinations, and "forgetfulness" of the large model will directly lead to a dead end for AI's penetration into the screenwriting industry.
However, APUS (Apasi), which is developing the screenwriting large model, does not think so. "We envision that what AI can do is to assist the work of screenwriters. Even more, we hope it can release the energy of screenwriters and return to creativity," the relevant staff of APUS told Workplace Bonus.
An AI screenwriter who can write a 70-point script requires at least 20,000 pieces of language materials and 500 episodes of scripts to train a preliminary model, and these data cannot be provided by screenwriters without 3 - 5 years of industry experience.
But the scoring criteria also vary from person to person. The 70 - point screenwriting large model defined by APUS is one that can give screenwriters some inspiration, and can also fill in the flesh and blood of the script according to the framework given by the screenwriter, and solve some basic problems. The biggest disadvantage of AI screenwriting is that the large model is not very strong in the ability of epiphany. In popular terms, it is not very "spiritual", but this is also the ability that users expect the large model to emerge.
"At present, the things that need to be improved and refined in the script still need to be done by humans," the relevant staff of APUS mentioned.
So, "Will AI replace screenwriters?" is currently a false proposition. This time, AI does not intend to replace anyone. It is more like an efficiency tool.
Recalling 2023, people were deeply trapped in the fear of being replaced by AI. But two years have passed, and AI has not replaced anything. It is just an efficiency tool to assist work. However, practitioners who do not know how to use AI do face a greater risk of competition from their peers.
Then why are there still so many job losses? "The current market situation does not seem to be caused by AI. Sometimes, AI even brings an increase," an investor said.
Making hundreds of formulaic short videos in one minute, drawing hundreds of images of the same style, and writing dozens of articles with the same overall structure... This is the efficiency improvement brought by AI. In the past two years, AI has won by quantity, but has not climbed the content field.
It seems that what AI makes difficult is not content creators, but those who urgently need to absorb content.
In this view, Facing the AI wave that is sweeping the screenwriting industry, it seems that the producers, rather than the screenwriters, should be the most headache.
What producers need to do is to accurately find the right one from thousands of scripts. However, in the script trial stage empowered by AI, the script quality of screenwriters is becoming more and more similar. Producers' previous experience of "panning for gold" from numerous scripts no longer works. How to accurately screen and discover truly potential works has become a problem in front of them.
● Picture source from Doubao, with the prompt word being the producer scratching his head because he can't find a suitable script
Producer Zhou Huike told Workplace Bonus, "When the content world is filled with AI, vibrant stories, tactile creativity, and resonant points of the same frequency will become the most scarce things."
The Butterfly Effect: AI Brings an Increase in the Overall Demand for Content Talent
In addition to screenwriters, AI has appeared in all aspects of short drama content creation.
Creators are the first to perceive the changes that AI can bring. Xi Xi, who has more than ten years of work experience in the traditional advertising industry, has paid early attention to the text - to - video software Sora. As soon as the domestic product "Jimeng" was released, he began to try it. "AI provides me with an opportunity to turn my ideas into reality." Without the need for more manual labor, with the help of AI's image generation, one person can become an advertising team.
Not only that, Wang Tianhai, the CEO of the hit drama film and television technology company, also mentioned that his 11 - person AI team only took about a month to produce 40 episodes of anime short dramas. "This includes a major rework," Wang Tianhai introduced. According to Workplace Bonus, a traditional anime production company's 10 - person team can only produce about half the amount (20 episodes) in a month.
● The anime short drama that the hit drama film and television technology company is preparing to launch during the Spring Festival
On the content creation side, video workers often follow such a workflow: first, turn ideas into pictures through Midjourney (or SD), and then put the pictures into video generation software such as Sora, Keling, and Jimeng to produce video content. The video needs to be continuously adjusted to obtain the video content that meets one's expectations. This also provides industry insiders with a new idea - when the shooting picture is flawed, they will also consider using AI to modify it.
In addition to image generation and adjustment, AI also provides more possibilities in the upstream and downstream of short dramas. Wang Tianhai told Workplace Bonus that the application of AI in the upstream and downstream of short drama production also shows more uses: virtual scenes, sound effects, actors, publicity and distribution, etc. "Short dramas are very cost - conscious. If AI can do well, the number of manual workers can be reduced, and the cost can be saved. Why not do it?"
A short drama company with a full - process AI has a monthly recharge cost of about 10,000 yuan on AI software (adjusted according to business). And the labor cost that AI can save for the company is as high as 50% [3]. For individual creators, the cost of generating AI videos is even lower. According to Xi Xi's personal experience, platforms such as Jimeng and Keling will select creators with high activity and high - quality works. Those who pass the selection, as super creators, have a large number of free generation times every month, which is equivalent to zero - cost creation.
In terms of virtual scenes, AI can also help short drama companies save tens of thousands of yuan in costs. The short drama "Qi Liang Ren Sheng" launched on Douyin at the beginning of the year, 60% of the content is produced using virtual production methods. The characters in the drama can "travel" to different crime scenes, leading the audience to quickly switch in various scenes. Among them, large - scale landscapes such as cable cars and city landscapes in the drama are all produced through XR virtual production technology.
The use of digital humans has broken the traditional rules of using actors in short dramas. In the traditional short drama industry, actors are usually paid on a daily basis, and the