#btw this is not intended to be an attack but a firm correction of what i parse as a well-intended misrepresentation of what is possible

[dalniente]

Of course it isn't "designed to rework" anything. Generative AI is generative. It generates its own new output based on user inputs and whatever parameters and restrictions you (JSTOR) build into your tool-- but it generates that new output in ways humans can understand by training on & pulling from vast datasets of stolen information. It builds sandcastles all its own, every single grain of which was stolen from a person. There is no ethical workaround for this. We were not asked, we were not paid; neither OpenAI nor their contemporaries gave us the opportunity to have our words and works excluded from their datasets.

Generative AI is trained on hundreds of billions of points of scraped data encoded into "tokens" comprehensible to its algorithms. To give you an idea of scale, GPT-3's text datasets included 3 billion tokens from Wikipedia. 86 billion came from three other sources, and a whopping 410 billion tokens came from the 501(c)(3) nonprofit CommonCrawl, which completes Nutch crawls on a regular basis. I believe that until recently, the petabytes of data in CommonCrawl's sets and archives (which are available to the public despite containing copyrighted work, under fair use claims) were mostly used for research, corporate datamining, and what I'll call "targeted" app development-- people using specific subsets of the data. GPT-3 and GPT-3.5's training data, however, is all-encompassing, and it is being used to generate enormous profit across multiple industries for OpenAI Global. You, JSTOR, are not profiting directly off of this-- I assume-- but you're using GPT3.5 Turbo at least in part, so OpenAI very definitely is.

(Anyone remotely familiar with fair use should immediately see the problem here. Not with JSTOR specifically, to be clear. But overall.)

I understand the attractive opportunities arising as this technology develops. I understand JSTOR'S need to stay competitive. I understand perfectly why JSTOR is developing this tool, and based on a brief review of your FAQ, it seems like it is designed to work with a minimum of hallucinations. If I'm reading between the lines correctly, you're basically trying to make this into the conversational search engine many people already believe LLMs to be. Well done. You also reassure that you will not replace humans. Good!

But that's only part of the puzzle. The use of generative AI at this time is unethical at its core. You do seem committed to using AI responsibly, as you claim...but to use something like this ethically simply is not possible using current tools and methods. The data in the sets upon which OpenAI's tools are built and trained does not belong to them; to use it for profit is fundamentally unethical. I am sorry.

I would remove "ethical" from future assurances. Replace it with "conscientious," perhaps. Or "as ethically as possible." I do not know that I would call that better, since it's a fundamentally empty reassurance, but at least it would no longer be a factual impossibility.

Like I said, I fully understand why you're developing this tool. I'm not going to tell you you shouldn't. But I will tell you flat-out: You can "commit" as hard as you like to building your sandcastle ethically. You won't make the sand any less stolen.

Why is JSTOR using AI? AI is deeply environmentally harmful and steals from creatives and academics.

Thanks for your question. We recognize the potential harm that AI can pose to the environment, creatives, and academics. We also recognize that AI tools, beyond our own, are emerging at a rapid rate inside and outside of academia.

We're committed to leveraging AI responsibly and ethically, ensuring it enhances, rather than replaces, human effort in research and education. Our use of AI aims to provide credible, scholarly support to our users, helping them engage more effectively with complex content. At this point, our tool isn't designed to rework content belonging to creatives and academics. It's designed to allow researchers to ask direct questions and deepen their understanding of complex texts.

Our approach here is a cautious one, mindful of ethical and environmental concerns, and we're dedicated to ongoing dialogue with our community to ensure our AI initiatives align with our core values and the needs of our users. Engagement and insight from the community, positive or negative, helps us learn how we might improve our approach. In this way, we hope to lead by example for responsible AI use.

For more details, please see our Generative AI FAQ.

[eatingfireflies]

11 Questions WIP tag meme:

Tagged by @chessanator​! Thanks for tagging!! :D

Tagging @greaseonmymouth​, @hardcoreprince​, @amberlyinviolet​ (i missed you!! ;;;), @billyweird​, @zossie​. But you guys don’t have to ofc!

1. How long have you been writing?

Me and my cousins used to make little illustrated booklets when we were kids. I was around 6-7 years old (probably) because I remember making Sailor Moon comics as well and I was that age when Sailor Moon showed in my country. When I was a kid what I really wanted to do was do comics, but things happened. :”D

2. What are the major themes of your current wip(s)?

I’m trying to think back and most of my WIPs seem to be about life getting pear-shaped and the characters having to deal with that. Two of them will need the ‘Major Character Death’ warning at AO3 if I ever post them. 

3. What do you want people to take away from your story once they’ve read it?

Depends on what fic they read. :”D But if a person got something from my fics that I did not intend, then all the power to them. So long as it’s not ‘that writer sucks’ lol!

4. Would you be excited if people write fanfiction about your wip(s)?

I’m torn because I love sharing headcanons and trends, and I’d be SUPER flattered if anyone wrote fic based on any of mine but there’s also the worry re: accusations of plagiarising and stealing someone else’s ideas. So personally, I’d be excited. But in a more general level it’s a bit of a thornier issue. 

And I know this is an issue with trad pub/canon vs fanfic on a bigger scale (like how some traditionally published writers won’t read fics based on their own work no matter how supportive they are of transformative work because of the whole plagiarism issue (accidental or otherwise)) but yeah.    

5. What’s your go-to writing beverage?

Water lol. I don’t need to drink anything to get going, but I do need cigarettes to function properly. Pls don’t start smoking. 

6. Who is your favorite oc? Tell me about them!

I’m going to talk about an OC from Travelling in company--a Junpei + Phi detective bros fic. So there will be spoilers.

But yeah I really like Hirata Ichirou. He’s the Mr Body in this fic: the character who was written to die so we can investigate what happened. At first I didn’t really know exactly what happened to him (lmao I knew it was cult-related but also personal but I didn’t decide until I was halfway done with the fic) and he was simply a nameless chinpira, but as the fic went on I realised I knew exactly what Hirata died of and why. I grew really fond of the guy. 

He likes to eat parfaits with his childhood friend, I mean. Come on.  

In actual WIPs uuuh. I don’t generally write OCs? Sorry folks. :”D

7. Do you feel that mistakes are important learning tools in the writing journey?

Not if I don’t catch the mistakes and correct them. One of the most important learning tools imo is reading books. Not fics by other people (which btw are really good and writers like franticbabbles taught me a lot about writing) but traditionally published novels. See how its done by someone who (hopefully) has a professional editor to ‘curb their spur’, decide what works for you and what doesn’t. I find that aside from having a beta, it helps you catch mistakes and decide whether you want to correct them or not. 

As Neil Gaiman said--and I paraphrase--your style are the things that you get ‘wrong’. But it defo helps to know the rules before you break them.

8. Rank your ocs by their capability in a footchase (either running after or from smth, your choice)

Here’s a bunch from Travelling in company because that fic has yakuza in it.

1, Honda Mitsuo/Wakaisha #1: very competent man. Can probably kill you with a paperclip

2, Mishima: not incredibly physically active but he has the brains and the resources to make it so he doesn’t have to run very far if he has absolutely no choice in the matter

3, Hirata Ichirou: okay dude. A chinpira and has probably run away from the police/other yakuza members before. Knows shortcuts and dead ends.

4, Satou Keiichi: more buff than fast on his feet. You can trust him to protect you, but not so much when it comes to running after escaping attackers.

Bonus from Kentucky for Christmas:

5, Hugh Grant (not the actor but a detective who insisted he looked like the actor): not very useful. Fell down a building while on the run from gangsters. 1/10 won’t recommend.  

9. Does your wip have romance? tell me about it!! if not tell me about a friendship/important relationship in your wip!!

Oh all the romance. I only write romance fics, didn’t you know? :”D All those mysteries and crimes are just frills for the romance and bonding. 

Right now I’m in the middle of writing a scene where Aoi and Junpei bond over Gunplas. 

10. Do you believe in the advice kill your darlings?

Yes. I have a ‘rags box’ which is a gdocs file where I throw all the lines of dialogue and narrative I had to cut from fics. Hoping that someday, somewhere, I can recycle them in some form. Don’t let them go entirely, they might still be useful for another piece if not the one you’re working on. 

Another thing I learned from this guy who taught Creative Writing in my first language was be mindful of your POV. In an exercise, he made us write about the emotions of an old woman going to a shopping mall right after the death of her husband. I was raring to go. I wrote about this old woman grabbing her daughter’s elbow tightly because she felt she was going to float all the way to the domed ceiling of the mall--that’s how light she felt. 

Teacher: Is that the old woman talking, or you?

Me: How do you mean?

Teacher: That literary~ language, is that her or you? 

Me: It was me.

Teacher: Great, rewrite that. 

Yeah, sometimes we have to kill our darlings, our floating into the ceiling moment, to present the experiences of a certain character fully. And that’s fine. 

(It’s also fine if you don’t want to. It’s just something that works for me.) 

11. Do you prefer plotting or worldbuilding? Why?

Worldbuilding. I’m so bad at coming up with events is why :”D I literally need prompts. But once I have a firm grasp on the world the characters are inhabiting I get more ideas on what they could do and what could happen around them.