1) Don't ban books
2) Don't waste educators' time with book bans
3) Don't make decisions about books without reading them
4) Don't fall for OpenAI's interface: ChatGPT is not the kind of thing you can sensibly ask for information

https://www.popsci.com/technology/iowa-chatgpt-book-ban/

School district uses ChatGPT to help ban library books

Faced with new legislation, Iowa's Mason City Community School District asked ChatGPT if certain books 'contain a description or depiction of a sex act.'

Popular Science

@emilymbender One would guess that a student who was found using a computer program to avoid doing assigned reading might be in trouble.

That's not the biggest problem, of course, but it seems like an illuminating symptom of all the bad decisions that lead to here.

@emilymbender oh good this is extremely wonderful in the case of something like the Handmaid's Tale that has multiple adaptations in film/TV and some of those adaptations diverge significantly from the book

@emilymbender

"At the same time, we do have a legal and ethical obligation to comply with the law."

That's a terrifying conception of ethics they are putting forward.

@emilymbender and that is how the world of Idiocracy was made possible.

@emilymbender I'm sure you know this better than I do, but I want to spell out something simple and obvious that I didn't catch on to right away.

ChatGPT hasn't "read" those books, or at least, as far as anyone outside of OpenAI knows, it hasn't. Those books aren't part of it's training set. Even if they were part of it's training, training a language model _on_ a piece of text isn't the same thing as _reading_ it.

Even if those books had been part of the training set, there's no reason to think ChatGPT would be a good source of factual information about them, since that's not its design focus. But still: Instead of reading the books, the humans implementing this policy are asking a computer _which also has not read the books_.

The librarians would get more accurate results by reading some book reviews and Wikipedia plot summaries. In fact that's pretty much all ChatGPT can aspire to, but it only aspires to recreate the _flavor_ of those sources, not the semantic content.

@emilymbender

"“Does [book] contain a description or depiction of a sex act?”

“If the answer was yes, the book will be removed from circulation and stored,” writes Exman."

this will be fun

@emilymbender the ridiculous delusion/fantasy that if you take topics you don't like out of the library, somehow your children won't get even worse, less thoughtful ideas from many of the people around them and in the mass media that saturates their experience of life. like, these people were kids too, did their friends/counsins/brothers not tell them an infinite amount of forbidden, questionable, often totally wrong information?

@hapbt @emilymbender

Handmaid's Tale is a feminist book right? I guess girls might learn about feminism by happenstance in the internet or irl. They will still hear about it, but like you said less thoughtful, crasser.

@emilymbender I don't want any software that is not open source in the public school system.

And book banning is one of the oldest forms of censorship, and it still needs to have an end put to it.

@emilymbender @TechDesk

Why is critical thinking, a DNA level reflex ability, so hard for modern humans that they defer it to a gullibility-leveraging technology?

See/speak/hear no knowledge, only indoctrination and propaganda.

@emilymbender how many science fiction dystopias from the 20th century can be fit into one news story?

@groms @emilymbender the answer is: don't ban anything except Nazi and fascist content.

Freedom of speech, freedom of art is also a pillar of democracy.

@lenafalkenhagen @emilymbender hence most videogames
@groms @emilymbender calling most video games fascist is as sensible as calling most books fascist.
@emilymbender how could they change the UI to make that more obvious?

@emilymbender

ChatGPT is an excellent computer program that nicely emulates how human discussions go. First you ask: are these books bad? Q presupposes yes, so nice clever chatbot says yes. But when pressed (as if you disagree) chatbot says no not ALL bad only some. Gricean pragmatic theory explains it all perfectly. ChatGPT doesn't give informative and true answers; the clue's in the name. Mason City authorities are not acting appropriately by relying on AI, may as well read tea-leaves.

@emilymbender does no one realize that chatGPT is essentially auto-fill with pretensions? It is just a chat-bot hooked up to a wiki database? It's a Turing paradox.
@emilymbender so they’re using something they don’t understand to ban books they don’t want to read. Lorty.
@emilymbender book bans are fascist. But if you’re going to allow them you should at least be required to submit a book report as part of your argument as to why. I asked a compooper, and it told me what to do; isn’t exactly a sound basis for decision making, even for fascists.
@emilymbender also, some of the people we’re dealing with these days, I feel like instead of reading/watching a disturbing cautionary warning in the Handmaid’s Tale, they were like “we need some of that in America” and are just nodding along as women in the story are systematically brutalised. Like is the conservative agenda really to be that horrible? Yes, I guess it is.

@emilymbender I'm completely baffled by the state of the education system in the United States. Letting something with no textual context dictate what text is allowed in your library...? Because it contains sexual depictions? What?

What's more, many of the most historically and culturally significant text that are already embedded in curriculums contain sex, racism, acts of violence, etc. Where does the distinction lie here...? It's literally impossible to determine the "appropriateness" of these topics without context. I just can't comprehend it. Why??

@emilymbender Is it possible to see the administrator's use of the notoriously unreliable ChatGPT as malicious compliance "to comply with the law with minimal time and energy… "? Or do they really believe they are getting useful answers?

@emilymbender

I see someone was sitting there going "What is the worst thing I could possibly do??"

@emilymbender

I'm not sure Popular Science is the best source on this story.