

New App Creates Short Episodes of South Park Using Generative AI
AI and Data Science Newsposted by ODSC Team July 24, 2023 ODSC Team

A new app called Showrunner AI allows users to create short episodes of South Park using generative AI. Trey Parker and Matt Stone are no strangers to AI as the pair earlier this year invested heavily into a deep fake VFX studio. Though they have nothing to do with the app, it allows users to create short episodes only using prompts.
Created by Fable Studio, the program allows users to take control of their South Park-inspired episode with assets based on the famous tv show. But the purpose of the app isn’t to just create fan episodes for South Park but to show off the technology.
Of the app and association with the program, Fable Studio stated that “South Park” imitations are strictly to show off their product’s capabilities, and Showrunner AI won’t be generating “. The abstract states “In this work we present our approach to generating high-quality episodic content for IP’s (Intellectual Property) using large language models (LLMs), custom state-of-the-art diffusion models, and our multi-argent simulation for contextualization, story progression, and behavioral control.”
The abstract also mentions a scenario where viewers who became upset with the ending of the popular show Game of Thrones use AI to create a more fitting ending. Though the app currently allows for the AI generation of episodes, the overall goal of the app is to allow users to incubate worlds of AI personas. From the website, The Simulation, the stated goal of their work is to help humans better understand themselves through AI.
But all of this comes in the midst of the Hollywood writer’s strike that was partially sparked by fears of AI pushing humans out of the writer’s room. Fable Studio, it’s no stranger to tinsel town as it got its start creating VR films and has won two Primetime Emmys.
So far neither Paramount or Parker and Stone have reacted publicly to the app. You can watch their demo reel of the app below: