Originally published at: NaNoWriMo endorses AI, faces backlash, and struggles to clarify - Boing Boing
…
5 Likes
If you’re wondering why people who care so much about words would do such violence to perfectly legitimate and important ones like “ableist” and “classist”, you might also be interested to know that one of the sponsors of the festival is ProWritingAid dotcom, which now offers shiny new “AI” functionality.
26 Likes
It’s really disheartening to see how many organizations and individuals have wholeheartedly embraced AI for uses that are antithetical to their previous core mission of getting people to exercise their own brains and improve their own skills.
One example that my kids just pointed out to me was a video that Mark Rober did for Google, advertising the capabilities of Bard. In some ways his videos have already been going downhill a bit recently but they’ve always been at least somewhat creative and educational, and he’s got that whole CrunchLabs side business that purports to teach kids to “Think like Engineers,” so it’s sad to see him shilling for an AI company that’s whole deal is to do the thinking for you. In the Bard video he said that “having Bard do it for me is a win-win!”
15 Likes
Take a cab to do the marathon, it’s fine.
21 Likes
What I’m hearing them say is I should use an AI to produce a summary of any NaNoWriMo books so I don’t waste time reading all that generated slop.
Seems like I can skip both of those steps entirely and jump straight to the part where I simply ignore them.
15 Likes
Seems like anyone who wants to be thought of as an “author” should spend at least as much time writing a book as they expect others to spend reading it.
16 Likes
I always miss the boat on these things. My friend wrote a Markov chain program that would generate gibberish on IRC. Feed it some books and it would statistically generate sentences based on the words from the books.
What I wanted to do with it, was capture bits and pieces of what it said and paste them into something slightly more coherent. But I was hung up on getting the source text I wanted (LotR) and didn’t just run with many pages of logs I had already saved from everyone in the channel playing with it for a few days.
Is AI, specifically LLMs, different than a Markov chain? Mechanically yes, it’s like the difference between a manual and automatic transmission. But both are statistical weighing and need a fair amount of data to work. Markov chain kind of reaches a point where no amount of data makes it work better, but even a tiny amount of data can make it kinda of do what it is supposed to do. Where as LLMs have a much higher threashold at both the upper and lower end.
Can you use technology to create art? Obviously.
If you just set it on autopilot and give almost no human input is it art? Probably not, what is even expressed if a person isn’t there to actively decide anything about it.
Certainly something aesthetic is possible. Just like dripping sand randomly on a plate can be pleasing, it doesn’t mean anything until perhaps I say a prayer and dump the sand off the plate.
6 Likes
Back to my regular statement that I have used since Chat GPT was launched to the public whenever anyone no matter how much more eminent and powerful than I am suggests something “written” by an LLM - “If you can’t be bothered to write it, we’re not bothered reading it”.
Has shut up many a management type.
11 Likes
“questions around the use of AI ties to questions around privilege”
Yeah, let’s talk about privilege: how about how many resources are required to run the datacenters, and the impact they have on our already crappy ecology, and how the AI bubble is benefitting a small cadre of speculationists…
Fuck these guys
17 Likes
And don’t get me wrong, I can think of legitimate uses for AI. I used it, for example, for helping me “standardize” my writing, as a random prompt generator for NPC character bios ideas in RPG campaigns. Not good generator, but decent remixer.
I also see things like “machine vision” (which is machine learning, just not fancy AI LLM technobro wet dream), but also “machine hearing” (trying to figure out if the interlocutor is angry or happy or sad), automatic transcription… things that augment our capabilities, not stole our creativity.
But obviously we’re not talking about that.
13 Likes
Kids, don’t forget that it’s not plagiarism if you steal the greatest works of every previous writer, throw them into a blender, and pour it onto a page with your name on the top!
Disclaimer: this may not be how your teachers view the topic when grading your papers. This is not legal advice. IANAL. YMMV.
12 Likes
Or direct humans rather than augment them. Cory Doctorow calls the general concept “reverse centaurs”.
9 Likes
The curse of descriptivism strikes again.
NaNoWriMo are obviously using these words not in the sense of their dictionary definitions, but as their common usage as a “get out of criticism free” card.
13 Likes
Classist and ableist overtones…
Alternative point: no it doesn’t.
5 Likes
There are lots of ways to promote creative writing for people who have disadvantages that don’t involve letting a machine do it for them.
At this very moment the Paralympic Games are showcasing thousands of elite athletes who might not be able to compete in athletic events against other Olympians in traditional competitions. The solution isn’t to lower expectations, it is to provide a modified framework that allows every participant to showcase their talent and ability.
19 Likes
I think they tried to make a statement like “Use Grammarly if you have to!” (Grammarly is now banned at universities in our city because it uses AI).
They are all about “Do what you need to do to write the novel!”, and they certainly weren’t saying “Use AI to generate the novel and that’s just as good!” They were saying something like “Everyone should be able to write, so do what you need to do to write something because it’s good for you!”
But they said it poorly, and the general leap-on-the-bandwagon-ness of the internet meant it was widely publicized as “NANOWRIMO SAYS AI NOVELS ARE THE SAME AS HUMAN ONES” which isn’t really what they were saying.
It’s also regularly forgotten that there is no contest, judging, winner or any sort of output evaluation. It’s purely personal goals. So them saying “Do what you want to with whatever tools you want to” fits their model.
What a mess.
7 Likes
Plus, you have to pay to get access to the better levels of LLMs so allowing their use actually potentially biases the event against lower-income earners.
12 Likes
Gaiman once wrote a story many years ago, “Diseasemaker’s Croup”, with some sort of Markov chain tool; as the title implies, it imagined a writer somehow compelled to pepper his prose with increasingly convoluted ailments. Can’t recall reading someone else trying something similar.
5 Likes
In other words the NaNoWriMo crew actually aren’t good writers themselves, or perhaps they used an AI to generate their copy. Which comes to the same thing.
8 Likes
