no subject

[firecat]

The Nanowrimo peeps shouldn’t have referenced ableism, classism, and privilege in their recent post about AI.

But I don’t understand why a person would be categorically opposed to all uses of whatever-is-being-called-AI-this month in a creative pursuit.

(The majority of my use of AI has been getting suggestions for cat names, so that tells you how much you should pay for this opinion.)

There are ways to use AI, for example as a prompt or name generator, that are not “getting it to write your whole novel.” Why would someone object to such uses?

The Nanowrimo posts everyone is piling on already say that using AI to do the actual writing misses the point.
It’s a problem for creators that so far no copyright law covers what AIs can consume, but that’s a separate issue from whether they have legitimate uses.

I seem to be at variance with most of my opinion bubble about this issue. Feel free to tell me what I’m missing.

Context: https://www.404media.co/email/3d9698b2-8c2b-41e7-bea4-7a1ac6916159/

Comments

[ursula]

There are a bunch of concerns about the way that AI:

• depends on sources that were gathered without permission or compensation
  
• doesn't provide any way to cite or credit specific sources
  
• relies on human labor for tagging, classification, and moderation without fair compensation
  
• uses startling amounts of water (for cooling) and energy for each query.
  

Non-AI prompt and name generators don't have nearly as many associated environmental and ethical issues.

[shanaqui]

All of that, and my specific objection to NaNoWriMo is the coopting of social justice language to position people who have objections based on those valid concerns (about copyright, plagiarism, unpaid/poorly paid labour, the environment and others) as being ableist, classist, etc. It's a calculated move that speaks extremely poorly of everyone involved in crafting that statement.

I do know people who use (so-called) AI to help with their disabilities and I don't judge them for that, if it helps them, because better, more ethical tools may not exist or may not be available to them. But they're individuals, and they don't try to sweep the issues with AI under the rug as NaNoWriMo initially attempted to do. (My understanding is that they've now attempted to qualify their statement and acknowledge there are some issues.)

NaNoWriMo doesn't need to have a stance on people using AI for stuff like generating character names or prompts. Characters name generators, prompt generators, and various other non-AI tools to help people plan and write novels already exist. They've existed for a long time. NaNoWriMo didn't need to write a statement about whether people can use AI to replicate that kind of function, so (while I cannot read minds) To me it seems that NaNoWriMo is referring to much broader possible use of AI, where the AI output is expected to be a large proportion of the result.

There are ways to say "look, we don't police the tools people use, if they say they've written something, we believe them", which would have been a valid thing to say... this was not that.

[violsva]

I think there's a kind of hair-trigger effect. After a writer has been hearing about all of the terrible training of AI, and the terrible output, and flooding submissions, and that it's already putting writers and artists out of work and lowering their wages, and also it turns out it's destroying the environment, and so on and so forth, for nearly two years now, no one is giving any reference to AI the benefit of the doubt. (And not just writers!)

Given that AI is a completely meaningless marketing term, maybe they should. But it's reasonable that people are not willing to take the time to carefully separate out what AI they think might be okay and what isn't.

Also, a lot of people don't think the question of legitimate uses is actually separate from the issue of training datasets. Many people are not willing to use any output from a system they know was trained on unlicensed copyrighted works, even if they might be okay with the use itself.

[sabotabby]

Other people have covered it well.

AI is a tool for stealing art from mostly poor artists and funnelling it to the ultra-wealthy who are already trying to burn the world to death. It uses ungodly amounts of water. It's bad at doing the thing that it does, but it's just good enough to put humans out of work.

And also, it makes bad art. It makes art worse. There will be a whole generation of kids who think that this is the proper way to write and the proper way images should look, and the writers and artists that come out of this generation will be objectively worse. Not only does it have a self-devouring problem (AI trends towards mediocrity and feeds on itself, so the more AI content is out there, the worse AI content becomes) but people who lean on it begin to sound like it. It makes people sloppier thinkers. It's like someone has invented a tool for those Reddit guys who say "I have a great idea for a novel but I don't want to write, so if you ghostwrite it for me, we can split the profits 50-50."

Whatever good use cases exist for it, they are far outweighed by the ecological damage alone.

[minoanmiss]

Did you hear about the AI-written misinformation laden mushroom guides that landed at least one family in the hospital due to misidentification? I don't particular want to use a word tool that produces coherent sounding world salad, unless coherent sounding word salad is what I need (as for a disoriented character's dialogue).

Plus what everyone else said.

[lilacsigil]

Copyright law does indeed cover what AIs can consume and there are several major cases going on about that now. That's one of my ethical issues and other people have covered the rest.

The kind of AI that I do support is analytical AI, which is producing some interesting results in fields like astronomy and cancer imaging, rather than generative AI.

[firecat]

Thank you for sharing your views! I didn’t know about your item 3.

[firecat]

It's a calculated move

Thanks for your views! Fair enough. I haven’t been following Nanowrimo carefully enough to know this.

[firecat]

Thanks for your views and the links!

Also, a lot of people don't think the question of legitimate uses is actually separate from the issue of training datasets. Many people are not willing to use any output from a system they know was trained on unlicensed copyrighted works, even if they might be okay with the use itself.

Yeah. I think it’s fine to eschew tools marketed as AI for ethical reasons, and I want to see conversations about the ways it is causing harm. Or if people think Nanowrimo is making the statements they are in order to pander to their corporate supporters, I want to know that too.

But I see something else going on in a lot of the conversations about this.

[firecat]

Thanks for sharing your views!

I agree AI’s output is worse than that of humans, even humans with only a modicum of talent.

I don’t agree that it’s going to make humanity, or any particular generation of humans, stupider or less creative, although it will have that effect on some individuals. Those claims have been made about every new invention since we first began banging rocks together, and what usually happens instead is that people don’t stop being creative but use them to express creativity in new ways.

[silveradept]

NaNoWriMo walked back the initial comment, and then put out a couple of things that may have dug the hole deeper. They accused some objectors of bullying and virtue signaling, and then said this hue and cry is of the same sort as people dismissing romance writers as unserious, fanfic writers as plaigarists, or people hoping to hit it big as indies as fools, which to me seems an apples-and-oranges comparison, and then said their neutral stance was in service of maintaining a civil and inclusive community, and would everyone please kindly take their big discussions somewhere other than NaNoWriMo?

It wouldn't take much to read both of those statements as saying "We're saying sorry because you bullied us into doing it, now will you please take your concerns about ethics and your opinions about what tools to use somewhere else so we don't have to grapple with them?"

[firecat]

Thanks for your views!

I have heard about the mushroom guides. AI doesn’t know what truth or facts are, and that’s a good reason not to use AI for anything professional or anything where truth and facts matter. I don’t think using AI for personal stuff like a first draft of a novel is in that category though.

[firecat]

Thanks for your views!

Copyright law does indeed cover what AIs can consume and there are several major cases going on about that now.

Yes there are. I should have said the details are still being hashed out.

The kind of AI that I do support is analytical AI, which is producing some interesting results in fields like astronomy and cancer imaging, rather than generative AI.

Yes I’ve noticed that too.

[firecat]

If they’re digging themselves deeper that way, then I understand people getting pissed at them. Thanks for the links!

[minoanmiss]

Hmm. In which layer of writing do you work on character voice? I know from happy experience you're excellent at it.

Another thing AI fiction can't deliver, from what I've seen, is distinct character voice. This is one of the hardest and most rewarding parts of writing for me because how we say things influences what gets said and what doesn't, so I have to start with it in my first draft. I don't think I could go back and weave it into an AI-written first draft.

[violsva]

I generally feel people should take advantage of the fact that the internet makes it easy to cite your sources!

[firecat]

Thank you! I work on it in my first and second drafts usually.

Yes, AI would not be able to help with character voice beyond the most basic tropes (I assume it could do a reasonable hard boiled PI or Victorian gentleman, but haven’t tested that).

I don’t use AI for writing — I don’t need to because I had writing ability installed at a very early age. But I’m shit at drawing/painting and also don’t like to do it (even after I took a course and got a little better). So I’ve used AI to generate images for personal use. So I can imagine someone who wants to tell a story but isn’t happy with their level of skill in description, e.g., using AI to help with that. If they leaned too heavily on AI for that they wouldn’t get better at it. But maybe they don’t particularly want to.

[firecat]

That is an excellent point. Me too.