内容来源：https://www.theverge.com/ai-artificial-intelligence/762594/fable-showrunner-edwatch-saatchi-interview

内容总结：

【前沿科技动态】曾参与创办Oculus Story Studio并获得艾美奖的爱德华·萨奇，在虚拟现实热潮退却后转向生成式AI领域，其新创公司Fable推出的Showrunner平台近日获得亚马逊资金支持。该平台试图打造"AI版Netflix"，让用户通过文字指令生成定制化动画短片。

目前Showrunner依托Discord服务器运营，用户可选择角色和艺术风格，通过输入指令控制角色对话与场景互动。虽然生成内容存在动作僵硬、台词生硬等明显AI痕迹，但平台已能根据指令生成诸如"马斯克与奥特曼讨论 homelessness SaaS化"等个性化短片。

萨奇坦言，早年VR创业的失败让他重新思考娱乐产业的未来。他认为生成式AI的核心价值不在于降低成本，而在于创造"可交互的新型娱乐体验"。尽管当前作品质量远不及专业动画，但他设想未来能与迪士尼等公司合作，推出基于知名IP的授权模型，让用户创作《曼达洛人》等作品的衍生内容。

针对AI取代创作者的争议，萨奇强调Showrunner旨在补充而非取代传统制作，平台雇佣人类艺术家参与视觉资产开发。其商业模式计划向用户收取每月10-20美元费用，并与内容创作者分成。但业界质疑，这种"用户付费为好莱坞打工"的模式是否真能实现其宣称的"革命性突破"。

目前该平台仍处于早期阶段，其能否真正突破AI生成内容的质量瓶颈，尚需时间验证。

中文翻译：

作为Oculus Story Studio的联合创始人之一，爱德华·萨奇深知自诩革命性的新技术推广起来有多艰难。尽管该工作室的三部动画作品中有一部曾斩获艾美奖，但由于公众对VR电影普遍缺乏兴趣，Meta早在2017年就关闭了Oculus Story Studio。VR时代昙花一现，但萨奇对其新转型的生成式AI项目Showrunner充满信心——这个刚获得亚马逊资金注入的平台有望获得成功。

Showrunner旨在将用户转化为"AI版Netflix"的快乐内容提示师
Fable创始人爱德华·萨奇试图将好莱坞的AI转型游戏化——通过一次次提示交互实现。
（若通过The Verge链接购买商品，Vox Media可能获得佣金。详见道德声明。）

与其他专注于幕后部署生成式AI的娱乐公司不同，萨奇及其Fable团队开发Showrunner的初衷是让用户能根据个人偏好生成定制内容。目前该平台依托Discord服务器运行，用户可通过选择角色和艺术风格，并输入提示词来指定角色对话与环境互动方式，从而生成短动画视频。

当用户要求生成"埃隆·马斯克与萨姆·奥特曼在办公室茶水间讨论将 homelessness 改造成SaaS服务"的场景时，Showrunner会生成基本符合要求的片段。这些片段会匹配平台预设剧集的美学风格，比如糅合《硅谷》与《恶搞之家》风格的动画剧《Exit Valley》。AI生成的角色声音刻意模仿真实人物，但带着不自然的僵硬感，明显透露出机器自动生成的痕迹。

目前该服务免费，但Fable计划未来向用户收取10-20美元月费。虽然Showrunner目前仅能基于自有原创内容生成视频，但迪士尼等公司已表示有意向该平台授权IP。

萨奇坦言在Oculus时期过于自负，公司结局让他学会谦卑。这段经历让他重新思考消费者真正的娱乐需求，并确信生成式AI才是答案。"Meta收购Oculus后我们傲慢至极，在好莱坞四处炫耀时都说'你们完蛋了，我们要接管一切'。但最终我们对行业的影响为零，VR电影收入可能只有10美元。"

萨奇认为VR的核心问题在于让用户陷入被动观看与主动交互的矛盾状态。这种设计本意是打造电影与游戏的混合体验，但萨奇自身对VR电影的冷淡让他意识到这是条死胡同。

他对生成式AI的兴趣源于开发尼尔·盖曼《墙中之狼》VR改编版时遇到的技术瓶颈。团队希望主角露西能与玩家流畅对话，但当时只能实现预设对白。这个难题促使萨奇开始探索构建能与人类复杂互动的"数字生命"，并最终与OpenAI团队展开合作。虽然未能完全实现目标，但这次经历让他确信生成式AI能开创全新娱乐形式。

"我们让露西实现了视频对话功能，但很快意识到要赋予角色真实生命，就必须构建完整的世界观。角色不能是孤立存在的罐中大脑，他们需要家庭和生活背景。"

这种通过规则定义的沙盒虚拟环境来增强AI角色立体感的理念，最终促使Showrunner使用SHOW-1模型生成了一系列未经授权的《南方公园》剧集。虽然能模仿该剧画风和配乐，但无法复现其喜剧节奏和角色间基于真人表演的化学反应。这些山寨剧集不仅不好笑，更像是拙劣的同人作品。但对萨奇而言，这个实验证明了Showrunner具备成为内容生成平台的潜力。

萨奇像多数生成式AI创业者那样畅谈Showrunner，但其乐观态度与平台当前输出质量存在落差。他称之为"AI版Netflix"，相信通过海量用户输入优质提示词，终将产出媲美《辛普森一家》《亢奋》或《玩具总动员》的作品。他强调平台真正的吸引力在于能创造比传统影视更互动的娱乐体验。

"AI时代的《玩具总动员》不会是廉价动画电影，而是可交互体验。很多人认为生成式AI只是降本工具，但我们将其视为新媒介。电影诞生之初不是为了帮剧院省钱，而是具有颠覆性的媒介，需要多年探索。我觉得行业现在把生成式AI硬塞进电影的做法，恰恰扼杀了这种探索性。"

针对生成式AI可能导致创意行业失业的争议，萨奇给出了业界标准回应：Showrunner旨在补充而非取代传统娱乐制作。他认为单纯为省钱而拥抱这项技术的想法很可悲，并强调虽然平台基于多个大语言模型构建，但始终与人类艺术家合作开发视觉资产，"因为失去人性化表达就会明显缺失某些东西"。

"AI确实会削减工作岗位，但这正是我们对追逐廉价特效模式不感兴趣的原因。如果强大技术只能用来裁员，那意义何在？没人会因'听说这是AI版《玩具总动员》，制作成本超低'而买票观影。"

萨奇认为用户真正愿意付费的是基于授权IP生成场景的能力。虽然Showrunner现阶段主要生成Fable自有IP的粗糙短片，但公司最终希望与迪士尼等大厂合作开发品牌模型，让用户能生成《曼达洛人》等作品的衍生场景。这将"使人们能创造数百万新场景、数千剧集甚至自制电影"。

萨奇畅想的Showrunner运作模式类似《罗布乐思》和《堡垒之夜》——不是指建造或大逃杀玩法，而是鼓励玩家自创地图并分享的生态。不过与免费游戏不同，Fable明确要求用户付费使用服务。如果平台真能生成充满想象力、细节精美的艺术化世界，萨奇的愿景或许不会显得如此可疑且略带剥削性。但Fable目前兜售的，听起来又是一次试图用AI完成人类艺术家早已做得更好的尝试。

英文来源：

As one of the cofounders behind Oculus Story Studio, Edward Saatchi knows how hard it can be to sell people on new tech that bills itself as revolutionary. Even though Story Studio snagged an Emmy for one of its three animated features, a general lack of public interest in VR movies led Meta to shutter Oculus Story Studio back in 2017. The VR era has come and gone, but Saatchi is confident that Showrunner, his new pivot to generative AI that just received an influx of cash from Amazon, can succeed.
Showrunner wants to turn you into a happy little content prompter for the ‘Netflix of AI’
Fable founder Edward Saatchi aims to gamify Hollywood’s pivot to AI — one prompt at a time.
Showrunner wants to turn you into a happy little content prompter for the ‘Netflix of AI’
Fable founder Edward Saatchi aims to gamify Hollywood’s pivot to AI — one prompt at a time.
If you buy something from a Verge link, Vox Media may earn a commission. See our ethics statement.
Unlike a lot of other gen AI-centric entertainment outfits focused on deploying the technology in ways that audiences aren’t necessarily meant to see, Saatchi and his team at studio Fable developed Showrunner with the intention of people using the platform to generate content tailored to their specific desires. Currently, Showrunner lives on a Discord server where users can generate short animated videos by selecting characters and art styles from a list, and then writing prompts dictating what those characters say and how they interact with the environments around them.
After being told that you want to see Elon Musk and Sam Altman standing in an office break room and having a conversation about turning homelessness into a software as a service, Showrunner will generate a clip that mostly fits that description. Showrunner’s clips are all styled to match the aesthetics of one of the platform’s preset shows, like Exit Valley, a cartoon that appears to be a cross between Silicon Valley and Family Guy. The characters’ awkward, AI-generated voices are meant to sound like the real people they are based on. And they tend to be animated with an odd stiffness that makes it clear how much of Showrunner’s output is automated by machines rather than crafted by experienced human artists.
For now, the service is free, but Fable intends to start charging subscribers somewhere between $10–$20 per month at some point in the future. And while Showrunner is currently limited to generating output based on its own catalog of original programming, other studios like Disney have reportedly expressed interest in licensing their IP to the platform.
When I spoke with Saatchi recently, he admitted to being a bit too high on his own supply during his time with Oculus and deeply humbled when that version of the company ultimately came to an end. That whiplash left him reconsidering what consumers really want out of their entertainment, and it convinced him that the answers lie in gen AI.
”You have no idea how arrogant we were right after Meta acquired Oculus, but I remember being in meetings across Hollywood to show off our ideas, and we were just like, ‘You guys are done; we’re taking over,’” Saatchi told me. “But our net impact on the industry was zero in the end, and our revenue from VR movies was probably $10.”
To Saatchi’s mind, the big issue with VR was that it kept users in a kind of limbo where they were expected to be both passive and interactive depending on which scenes they were watching. Alternating between those two modes of engagement, Saatchi told me, was part of Oculus’ plan to make its projects feel like crosses between traditional movies and video games. But Saatchi’s own disinterest in watching VR movies was a clear sign to him that the technology was a dead end he should move on from in favor of something more dynamic.
Saatchi’s interest in gen AI was actually sparked by a technical roadblock he and his collaborators ran into while developing a VR adaptation of Neil Gaiman’s 2003 children’s book, Wolves in the Walls. In both tellings of the story, a young girl named Lucy lives in constant fear of the wolves living in the walls of her house, while her family insists that the creatures aren’t real. Saatchi and his team wanted their version of Lucy to be able to have fluent conversations with players / viewers as she guided them through the various rooms in her house. But the character was limited to reciting canned bits of dialogue rather than responding with context-specific speech.
This hurdle got Saatchi thinking more seriously about how he might be able to build Lucy as a complex “digital being” capable of having complicated interactions with people. That concept put Saatchi on a path to working with a team from OpenAI to see if it was possible. It wasn’t, not really. But the experience of building a slightly more robust Lucy character convinced Saatchi that generative AI could be the key to creating a new kind of entertainment experience.
“We made Lucy into a character that you can talk to and video chat with,” Saatchi said. “But what we quickly realized is that if you want to make a character truly live — which became our big goal — then you have to build a simulation of their world. They can’t just be a brain in a jar, like one character by themselves. They have to have a family, they have to have a life.”
The idea of building simulations — sandboxed virtual environments defined by specific rules — to make AI characters feel more multifaceted by giving them contexts to exist in is what led to Showrunner using its SHOW-1 model to produce a series of unlicensed South Park episodes.
Showrunner could approximate South Park’s visual style and musical cues, but it struggled to re-create the show’s comedic patter or the kind of chemistry between characters that, traditionally, is rooted in human actors’ performances. Also, the ersatz South Park just wasn’t funny, and it felt more like poorly written fanfiction than episodes of television that people might actually want to watch. But to Saatchi, the experiment demonstrated that Showrunner could be fashioned into a service — one dedicated to giving its users a way to prompt up “shows” of their own, one AI generated scene at a time.
Saatchi speaks about Showrunner the way many pro-gen AI founders do — with an optimistic enthusiasm that doesn’t exactly feel justified when you look at what the platform is currently capable of churning out. He sees it as the “Netflix of AI” and thinks that, with enough users writing the right prompts, it could produce something comparable to The Simpsons, Euphoria, or Toy Story. But Saatchi also believes the real appeal to Showrunner is its ability to create entertainment that’s more interactive than traditional films and shows.
“We think the Toy Story of AI isn’t going to be a cheaply produced animated movie, it’s going to be something that’s playable,” Saatchi told me. “Most people feel that generative AI is a tool to make the same, but cheaper, and we’re trying to say it’s a new kind of medium. Cinema was not about saving theater owners money; it was highly disruptive and took years to explore as a medium. I feel like the industry is kind of cutting off that exploratory element with generative AI by just shoving it into movies.”
When I brought up the ongoing conversation about gen AI’s potential to put people in creative fields out of work, Saatchi said what almost everyone in his position says — that he sees Showrunner as a platform that’s meant to supplement traditionally produced entertainment rather than replace it. He told me that he finds the idea of studios embracing this kind of technology strictly for cost-saving reasons rather grim. Saatchi also stressed that, while Showrunner is built on a number of LLMs, the company works with human artists and animators to develop its visual assets “because something is just clearly lost without that.”
“I don’t think there’s any papering over the fact that AI is going to cut jobs, but that’s why we’re not very interested in the whole cheaper VFX paradigm that most other folks are going after,” Saatchi explained. “If all that we can do with such a powerful technology is just cut jobs, what was the point? Nobody’s gonna go to the cinema to say, ‘I heard this was the Toy Story of AI. I’ve really got to get my ticket because it’s so cool that they spent so little on this.’”
What Saatchi does think people will be willing to pay for is the ability to generate scenes based on licensed IP. Though Showrunner’s core use case right now is making short, unpolished clips based on Fable’s in-house properties, the company ultimately wants to partner with major studios like Disney to develop branded models that would allow, for example, you to prompt up scenes featuring characters from The Mandalorian. This would “give people a way to create millions of new scenes, thousands of episodes, or even their own movies,” Saatchi reasoned.
”Our idea would be that, instead of people getting excited about stormtroopers in ancient Rome, which is, like, a cheap concept, there’s a Star Wars model that 700 people have developed under Dave Filoni’s direction,” Saatchi said. “These models would have real characters and a world that could be explored through prompting, and you could also inadvertently trigger scenes within those worlds in a way that would make it feel as though you’re uncovering something unknown.”
A clip from Fable’s Everything Is Fine.
Throughout our conversation, Saatchi was insistent about Showrunner being a good thing and a revolutionary tool designed to give users a new way of engaging with media. But he agreed when I pointed out that the system he’s describing makes it sound like Showrunner would effectively turn its subscribers into unpaid employees working for some of Hollywood’s biggest and most powerful studios. Studios would own anything generated with Showrunner’s branded models trained on copyrighted IP, and users will eventually have to pay to use the service.
But Saatchi stressed that, while Showrunner definitely wants to work with companies like Disney, he is also interested in collaborating with smaller creators who would stand to benefit greatly from the company’s business model. An indie filmmaker could license their new project to Showrunner and subsequently be paid a portion of revenue share based on how many scenes people were generating with the model based on their movie. Saatchi could not give me a timeline on when Showrunner might start trying to establish those kinds of partnerships, but he was bullish about them being part of what makes the platform a boon to independent creators.
“This could create something where creators can earn money when people are emotionally connected enough to their work that they themselves want to make something with it,” Saatchi said. “Compare that to what creators earn just from people viewing their work online. Yes, there is a kind of ‘we’re all employees of Disney’ element, but from a moral point, I can’t think of a better way to do it.”
Listening to Saatchi describe what he wants Showrunner to become, it actually sounds a bit like Roblox and Fortnite. Not the building or battle royale of it all, but rather the way those games encourage players to create their own maps, share them, and get other people to do the same thing. The Roblox Corporation and Epic have both built platforms where being a consumer can also essentially mean being a worker — one whose labor serves only to contribute to the corporations’ bottom lines.
But whereas those games are free to play, Fable very much wants people paying upfront to use Showrunner. If Showrunner were truly capable of conjuring up imaginative, detailed worlds that felt like thoughtful works of art, Saatchi’s pitch might not sound so dubious and mildly exploitative on its face. But what Fable is shopping around right now sounds like yet another attempt at using AI to do something that human artists are already quite capable of doing much, much better.