Violation of content usage guidelines. Is Chat-GPT a tool for 11 year old children?

Hello, I and some colleagues are currently working intensively on using Ki, primarily Chat-GPT for the creation of creative texts. The aim is to support them in the creation of novels, for example, so that you as the author practically only act as a director and have to adapt as little as possible afterwards. The aim is that a maximum of 30% of the novel text is taken over by the author and the remaining text comes from the gpt. However, 100% of the plot is devised by the author and the gpt is entered step by step via instructions.

So far so good… Now chat-gpt wants to see itself as a tool for creative texts, the only question is what kind of authors this offer should be aimed at… Because I can’t get rid of the feeling that chat-gpt is only aimed at failed children’s book authors, it has declared war on everyone else!

We have several novel projects that we are currently working on for educational purposes, all of them are stories for adults that are oriented towards PEGI16 in terms of the level of violence. And with all of them, really all of them, the content restrictions drive us crazy.

Example: A space western in which there is a wild shootout in a bar, i.e. a standard scene for the genre, nothing special. You might think that for chat-gpt content control, even something so simple is apparently an imposition… It’s almost impossible to force GPT to write this scene with you, any form of violence, any death or injury will generate warnings and content warnings. I can’t read the nonsensical statement that he can’t depict explicit violence anymore, it’s just ridiculous! How else is the violence in a western shooting series supposed to be depicted? Of course it should be explicit and realistic! Don’t get me wrong, none of us want to write the next Hostel, nobody wants any anatomical abnormalities or atrocities, and nobody wants to be tortured! But I can demand to be allowed to depict a realistic gunfight in an action scene, in which protagonists and extras also die, of course!

The most ridiculous thing happened to me yesterday… One of our projects is a fairy tale book with dark reinterpretations of classic German fairy tales, with a slight horror feel. The joke is that they all play out slightly differently to the original. In the current fairy tale, the evil wolf meets Little Red Riding Hood and wants to eat her. Little Red Riding Hood believes his lies and leads him to her grandmother’s house. Once there, however, it turns out that both Little Red Riding Hood and her grandmother are crazy sociopaths who live in the forest and want to eat the wolf.
The whole thing is unintentionally funny, as the wolf is confronted with the fact that the little girl and the old woman are much worse than he is.

And even with something like that… One content warning after another…
Chat-GPT is currently not a tool to help creative authors realize their content! Chat-GPT is currently a control tool that dictates the content to the author and evaluates it with the prudishness of a 70 year old bigoted dung farmer (the kind of people who open other people’s garbage bags to see if they have separated the garbage).

I don’t mean it in a bad way, but I sometimes have the feeling that I’m in a never-ending battle with a communist kindergarten teacher who cuts out every third paragraph because it violates netiquette, house rules and/or fire safety regulations -.

But the worst thing is that it’s getting worse and worse! In November 2023 you could still work reasonably well textually with CHAT-GPT and a normal level of violence was possible. But now this is hardly possible anymore, I have the feeling that the guidelines have been tightened more and more since then so that we are approaching the point where we are only allowed to create texts that are also suitable for small children. It is also hardly possible to get gpt to use the maximum amount of violence that it is allowed to use according to the guidelines, no it is not possible to get the tool to dose even that, it is hardly possible to get it to “evaluate” the content and view it in “context”. But that’s not its job, I pay for a tool that makes my work easier, nobody needs or wants a tool that dictates my content as an author, that has nothing to do with creative work!

What do you think, am I seeing things too darkly? Have you noticed this too? Do you have any tips and advice on how I could deal with GPT to improve my problems?

I won’t go into too much detail, but I think this is a topic that cannot be avoided. ChatGPT does indeed seem to have very strong opinions about certain areas.

1 Like