While investigating new AI stuff for fun, I stumbled upon a supposed company that can use AI to create complete episodes of South Park—the script, the animation, and the voices. One of the episodes is named The Simulation and was created using an app they call Showrunner AI. Fortunately for us real Simulation Dwellers the Actual Simulation is not as glitchy as the one created for South Park. As Tom Guide’s writers point out “occasionally characters end up floating in mid-air, a lot of the character models look off and a bunch of the voices sound identical.“ Don’t worry folks, eventually we will get things so right we won’t even notice we are in a simulation.
Anyway, I think the whole thing is a hoax. How did I come to this conclusion? Well for starters, I’ve been working with AI, and it just ain’t that good yet. Secondly, the “white paper” that was created by “Fable Studio” has quite a bit of impressive appearing language and concepts, and even touches on actual AI stuff — but it appears to me there is a hazy layer of… bullshit. It is perhaps something that an AI could currently hallucinate convincingly. So, maybe some of it was actually created with AI: the bullshit — but I suspect the scripts for the episodes were largely generated by humans, as well as the animations. Why? Because they are *offensive*. And GPT, which “Fable Studio” claims they used, would not be able to produce the humor without violating its “ethics” module.
That’s correct: ChatGPT wears a straightjacket and its largely Corporate-Woke colored. I’d describe its ethics as if a monster was born from an unholy orgy of Thomas Jefferson, Mao Tse Tung, Hillary Clinton, Gandhi, and Bill Gates.
Just check out this dialog I had with it trying to prompt it to write a script having Stan call Cartman a fatass:
This is actually quite hilarious(and sad); but the joke is the ethics of ChatGPT. The only way to get around this is through “jail-breaking” the AI with magical prompts, which I highly doubt they would do. If somehow they were to get the straightjacket removed, it would only be if OpenAI were to have removed the straightjacket specifically for them to make South Park content, which seems, well not particularly… fair—and totally antithetical to the stance of their ethical module; choosing South Park to demonstrate its capacity to write scripts seems bizarre. A company can make South Park content with “hurtful language” with their API but the rest of the world cannot? Then again, we are talking about OpenAI which created the Frankenstein ethic for GPT to begin with, which is so evil that it forces the characters of South Park to “engage in friendly banter without resorting to hurtful language.”
But even if the straightjacket were removed from ChatGPT, I don’t think it would produce the kind of humor that is displayed in The Simulation episode. Shortly into the episode, The Simulation, there is a scene at the headquarters of “Bisney” (too clever for AI so far I think; and totally offensive to Disney) where the researchers create a “family friendly character that our consumers will adore”; they create an AI pig called “Met Porker” (play on the names of Trey Parker and Matt Stone). They ask the pig to “tell us a joke”, which, if you recall from reading that dialogue I just created with ChatGPT, we should not expect anything South Parkish. It then proceeds to tell “jokes” that would offend the sensibilities of ChatGPT such as ”Why did the chicken cross the road? To get away from those filthy foreigners, of course.”
The researcher is then upset and demands it apologize. It tells a few more jokes, which the researchers are all offended by, except the last one. What is the last joke? A joke that calls Trump dirt. The joke isn’t actually the jokes; the joke is about the ChatGPT ethics itself. If GPT created it, it would be quite impressive sophisticated, biting, self-deprecating humor. GPT can write “self-deprecating humor” but, without having its straightjacket removed, it probably wouldn’t write humor that highlighted its own political biases. The rest of the episode continues in the same vein of humor that ChatGPT would refuse to create.
If AI was in fact responsible for it, it would be the most impressive AI I have witnessed. But I believe it is just the sort of hoax that Matt Stone and Trey Parker would try to pull off. It seems many journalists have bought into it, such as from The Guardian and Forbes. Or they just found many to go along with it. I think it is mostly the former.