• Welcome! The TrekBBS is the number one place to chat about Star Trek with like-minded fans.
    If you are not already a member then please register an account and join in the discussion!

Should we allow for AI-generated fiction writing?

Someone on my fanfiction group on FB said earlier today that she likes to use Chat GPT to check her punctuation. :brickwall:

I told her to get a beta reader. A human one.
 
Which begs the question: If it's actually fair use, why are they trying so hard to hide it?

I imagine there are actual technical secrets there that companies wouldn't want out in the public. They are also in a competitive market place and don't want to help the competition out.
 
I imagine there are actual technical secrets there that companies wouldn't want out in the public. They are also in a competitive market place and don't want to help the competition out.

Mm. Okay. Technical secrets like which writers they're stealing from, I guess.
I fail to see how disclosing what copyrighted material they're using is a "technical secret".
 
Mm. Okay. Technical secrets like which writers they're stealing from, I guess.
I fail to see how disclosing what copyrighted material they're using is a "technical secret".

From your post…

AI companies that are developing “general purpose AI models,” such as language models, will also need to create and keep technical documentation showing how they built the model, how they respect copyright law, and publish a publicly available summary of what training data went into training the AI model.

Should we also get the library records/book purchases of every writer out there to make sure they haven’t borrowed something from something that they read?
 
Last edited:
From your post…



Should we also get the library records/book purchases of every writer out there to make sure they haven’t borrowed something from something that they read?

Technical documentation of how the built the model is exactly what I said. It's a legalistic way of stating "We're going to see whose copyrighted material you are using".

And you're acting like real people don't get hit with the plagiarism button all the time.
 
Technical documentation of how the built the model is exactly what I said. It's a legalistic way of stating "We're going to see whose copyrighted material you are using".

And you're acting like real people don't get hit with the plagiarism button all the time.

“How they built the model” would likely include trade secrets on how each company is creating the hardware/software for their version of AI.

A list of materials is fine, anything beyond that will never fly.

But, one would still have to prove that how AI learns is any different than how a human gains information and that their specific work is being used.
 
But, one would still have to prove that how AI learns is any different than how a human gains information and that their specific work is being used.

No one has to prove anything. It's already established as copyright law. At least in the EU. And it's being established as so in other courts as well.
 
“How they built the model” would likely include trade secrets on how each company is creating the hardware/software for their version of AI.

A list of materials is fine, anything beyond that will never fly.

But, one would still have to prove that how AI learns is any different than how a human gains information and that their specific work is being used.
How they built the model is, but the model by itself is blank and generic. The training data is what we're talking about and that should be public. ChatGPT sometimes can directly replicate its source material if you give it a prompt it was trained on, and can give the same exact answer to multiple people, both of which are copyright nightmares.
 
...ChatGPT sometimes can directly replicate its source material if you give it a prompt it was trained on, and can give the same exact answer to multiple people, both of which are copyright nightmares.

Use with caution.

I'm currently using YES lyrics as writing prompts. If you're getting stuck coming up with story ideas, there are more useful ways to come up with stuff than AI.

I'm not bashing AI. As a contract manager, I use it all the time. It saves me a lot of time in replying to business emails, developing policy and procedure documents, planning and training documents, Gantt charts, and definitely in writing contracts. Easily doubling my efficiency.

It can also get me into a lot of trouble if I don't carefully review what it produces for me and bend it to my will instead of allowing it to bend me to its algorithms.

rbs
 
Why? M$-Word has a built-in grammar / punctuation checker that's 90+% accurate.
I have no idea. I think she said English isn't her first language, so maybe that's a reason?

The problem with using anything commercially programmed is if you write a lot of fanfic that uses dialects that have non-standard grammar, MS Word is basically pointless. It's programmed for standard American English, which means that there are normal Canadian words that get flagged, never mind something that's written for a science fiction/fantasy-themed fanfic.
 
The problem with using anything commercially programmed is if you write a lot of fanfic that uses dialects that have non-standard grammar, MS Word is basically pointless. It's programmed for standard American English, which means that there are normal Canadian words that get flagged, never mind something that's written for a science fiction/fantasy-themed fanfic.
I have found that most of the writing in my scifi stories is standard. For the 1 to 5 percent of the made-up terms, sentence structures, punctuation emphasis that I add, I have a choice to either add the term to the user dictionary or ignore the computer's suggestion. That way, I still benefit from the service offered from a spell check and grammar check computer-driven program and have the flexibility of doing my own thing.

Sometimes, even for standard language use, the computer gets it wrong but that doesn't mean I turn it of and just live witmy all my mistakes. It means I remember that even though I have an efficient, and essentially free editor, I still have to read it through myself and pay attention. It is up to me as to weather I have used the word I meant to use our whether I let the computer be the author of my thoughts.

There is a great deal of value in accepting the help of a computer. I'm not going to give up on that value just because I don't like the idea that it could be misused or that it might suggests some words that I hadn't thought of or that a small percentage of the time it isn't right. Frankly, as an author, it's a non-issue to me. I'm writing my stuff. Someone may easily read my stuff and quote it without giving me credit by changing the context, or a word or two, and I would never know. It doesn't matter if it is a person or a computer. I put it out here to be read and hopefully appreciated. I'm just grateful the AI bots are not all over my writing with harsh criticism. No one needs a computer to also be a critic.

As a reader, if its a great story, I'll read it, if it's not a great story, well, I already mentioned how there are enough critics on the Web. I'll read another story.

-Will
 
Last edited:
(sic)

I'm far from certain you didn't do that on purpose. It's far too ironically case in point...
Actually, that was not on purpose, but it does work. And, three words before your quote starts, it should have been the word 'off' not 'of'.

-Will

P.S.: This is me not taking credit for something the computer wrote, but using my artistic discretion to leave it.
 
Last edited:
How they built the model is, but the model by itself is blank and generic. The training data is what we're talking about and that should be public. ChatGPT sometimes can directly replicate its source material if you give it a prompt it was trained on, and can give the same exact answer to multiple people, both of which are copyright nightmares.

My thing is though... so... we have copyright laws.

I don't see why we need to ban a tool when we already have processes and laws to deal with copyright issues. There's really no functional difference between my human brain copying something and a computer copying something.

Just... apply copyright law.
 
If you are not already a member then please register an account and join in the discussion!

Sign up / Register


Back
Top