This message was deleted by the user at 03:46, Tue 05 Mar.

I used it once to generate some myth ideas for a character I was having trouble with, and it did alright -- it gave me a stating point. Which is what you generally get out of any sort of random generator.

However, I would advise against trying to develop anything mechanical with ChatGPT. It doesn't do math -- it's a sophisticated chatbot. So anything you get from it will be made from it Frankensteining together words in a sequence that looks "about right." It doesn't understand the words it outputs, so it has no check or balances. You will get something that superficially looks like useful rules, but that's just artifice, and there's no guarantee you'll get anything that actually works in practice.

This message was deleted by the user at 03:46, Tue 05 Mar.

Generative AI doesn't "understand" anything. I'll grant this particular program sounds like it has an integrated calculator in it, which can be helpful, but "doing advanced math" isn't the same things as "creating fun mechanics."

With the right prompts, you can get a ton of good results, but there's plenty of limitations. It doesn't have a "mind," "consciousness," or "understanding" in the same way a human might, no matter how humanlike its programing makes it behave.

I'm going to use an analogy, the Chinese Room thought experiment. Imagine a native English speaker who knows no Chinese locked in a room full of boxes of Chinese symbols (a data base) together with a book of instructions for manipulating the symbols (the program). Imagine that people outside the room send in other Chinese symbols which, unknown to the person in the room, are questions in Chinese (the input). And imagine that by following the instructions in the program the man in the room is able to pass out Chinese symbols which are correct answers to the questions (the output). The program enables the person in the room to pass the Turing Test for understanding Chinese but he does not understand a word of Chinese.

So it is with AI. It doesn't understand "context," it's just been programmed with enough data sets to output relevant results. It's following your inputs, and producing results (based on its dataset) in line with what you expect to see, but it's not actually creating anything. It's modeling. It's a useful tool, but like all tools, has limitations that a human mind won't have.

Now, there's a lot of philosophical back-and-forth about whether these things COULD achieve true understanding, but we're not there yet with the tools right now.

This message was deleted by the user at 03:47, Tue 05 Mar.

I don't doubt it could come up with "believable" backstories. That's not what's at issue here.



It can synthesize a step-by-step process from details in its database. If there are enough instances of similar steps in the same order it can reference, it might give you something that works, if that information comes from accurate sources. On the other hand, it might be mix-and-matching steps from inaccurate sources and misinformation. It has no way of telling if those "step-by-step" instructions actually fix the problem, nor whether its instructions are misleading in a catastrophic way thanks to miswording.

You might get the results you want following the steps it outlines. You might not. And the AI won't care either way. It doesn't know the difference between outcomes that fail because you followed the instructions wrong vs. outcomes that fail because it gave you bad information in the first place.



Would you? I'm going to bet it has more PDFs in its database to draw on than just the one you uploaded, and rephrasing things just means it's searching a different part of its database for those words in sequence. Like the guy in the Chinese room, it can find form and match it, but it doesn't necessarily comprehend the spirit of things.



Your low opinion of "most people" isn't helping your case.



I wish you would stop assuming I haven't. And I'd like you to stop trying to evangelize me, thank you very much.



I have, in fact, done similar things with ChatGPT before. If you prompt it right, you can get lots of interesting things. However, you are still doing the actual creative work. You are the one feeding it responses to get the outcomes you want. It can certainly help get you past creative blocks, but it's not a partner -- it's a tool. And I get it, I love participatory world design myself.

This is not that. This is a sophisticated random generator. And it should be treated as such.
This message was last edited by a moderator, as it was moot, at 18:10, Mon 04 Mar.

I have used it in a brainstorming capacity, and it's very effective at coming up with ideas quickly, be that in lists or for designing structured scenarios. It can effectively 'riff' on inputs that I've provided, and assist with refinement of my own ideas by coming up with options, then I can choose between them, and use it for further refinement and design. It takes a lot of time out of things I could do myself, leaving the fun parts for me.

This message was deleted by the user at 03:47, Tue 05 Mar.

The better you write your questions and provide examples, the better the results you will get.

People that like using the Game Master Emulator will love using AI once they get the hang of it.
It is yet to be interactive in that you have to prompt it for responses. It doesn't currently offer any unsolicited responses so, doesn't feel like a person. Yet.

AI is effectively banned from my games.  If a player can't be bothered writing their own posts then I don't want them in my games.  To me AI is the opposite alignment to RPG and never the twain shall meet.