Care & Feeding: My son can't do this.

[minoanmiss]

Full title: My Kid’s Teacher Has Assigned My Son a Project That Will No Doubt Turn Him Into an Incel


Dear Care and Feeding,

I’ve just been informed of an assignment my son “Patrick” was given in middle school. He’s to select a topic from a pre-approved list, ask ChatGPT to write an essay about it, submit that essay to the teacher, and then fact-check the essay, looking for things the AI got wrong. I do not want Patrick doing this assignment. ChatGPT is an irreducibly sexist device that reads and then regurgitates garbage that pushes impressionable young men into misogyny and inceldom. He is absolutely banned from using AI of any sort in our household, and if I can keep him from ever interacting with it until he’s 18, I consider it the least I can do for him as a parent. Only now the school is mandating he use it. Is there any way I can get the teacher to change his mind about this assignment? Surely it can’t be that hard to just produce a bad essay and have him fact-check that, right?

—Can’t We Just Protect the Kids from the Robots?


Care and Feeding is Slate’s parenting advice column. Have a question for Care and Feeding? Submit it here.

Dear Care and Feeding,

I’ve just been informed of an assignment my son “Patrick” was given in middle school. He’s to select a topic from a pre-approved list, ask ChatGPT to write an essay about it, submit that essay to the teacher, and then fact-check the essay, looking for things the AI got wrong. I do not want Patrick doing this assignment. ChatGPT is an irreducibly sexist device that reads and then regurgitates garbage that pushes impressionable young men into misogyny and inceldom. He is absolutely banned from using AI of any sort in our household, and if I can keep him from ever interacting with it until he’s 18, I consider it the least I can do for him as a parent. Only now the school is mandating he use it. Is there any way I can get the teacher to change his mind about this assignment? Surely it can’t be that hard to just produce a bad essay and have him fact-check that, right?

—Can’t We Just Protect the Kids from the Robots?

Dear Can’t We,

Advertisement

I grant you, AI is spooky. But it is here to stay. And for all its downsides (among them: being so easy to use for cheating, deception, and creating false narratives and “news” [absolutely not an inclusive list]), it is not responsible for creating incels. I promise you, there is no need to exaggerate AI’s potential harm.

Patrick’s teacher is smart—creating an opportunity for his students to see for themselves what a crappy job AI does at the thing kids are most likely to use it for is clever. Pretending it doesn’t exist, forbidding its use at home, and imagining that your son will never be tempted to use it to write a paper for him is not going to protect him, even if it makes you feel better. But an exercise in discovering its limitations will help to make him skeptical about it; it will make him smarter about it. That’s exactly what you want. And it’s not what your proposed substitute assignment will do.

Comments

[dissectionist]

Oh goodness. On one hand, I applaud a parent who wants to prevent their kid from being radicalized into hatred. But unless you’ve got a rural farm with no internet access and no TV, and nobody else ever visits, it’s impossible to prevent your kid from being exposed. And even with that rural, disconnected farm, at some point your kid is going to leave once they’re grown.

So what do you do? Expose them with guidance. Watch videos about online radicalization together (I strongly recommend The PewDiePipeline video, available on Youtube) so they know what to look for and how to recognize it happening in their friends. Discuss what incels believe and why it’s not actually true. Discuss misogyny and why what they propose is inaccurate, and the real reasons behind something when something they say looks accurate on the surface (like they’ll claim that family courts are biased against men because women get the kids way more often, but when you dig deeper, you find that in areas where women do get custody way more often, it’s generally that many dads don’t ask for custody because they don’t want it).

If your kid has no exposure, they don’t learn the skills of recognizing subtle bullshit or radicalization. They don’t learn the skill of investigating oppressive claims. They don’t learn how to protect themselves by knowing the reality first. LW, you’re advocating for the oppression equivalent of abstinence-only sex ed here, and that just doesn’t work. Give your kid the skills to have the info, and get the info in the future, that will allow them to recognize oppressive content and not be duped by it.

[topaz_eyes]

First, LW should ask for the pre-approved topics list to see what kind of misinformation may be involved in generating the essay. I do agree with C&F that this assignment does exactly what LW wants their son to learn about AI: garbage in, garbage out. However, I don't know if middle school is the best age to do an assignment like this; high school is probably better.

Unfortunately AI is also used on Google's search engine. Fact-checking the AI essay by another AI, one rigged to display the most popular results rather than the correct ones, compromises the assignment. Unless the teacher is aware of this and has guidelines on how to use internet search engines appropriately (eg use only the high-quality information sites, check with multiple sources, etc).

[ethelmay]

One would hope the teacher tells them how to bypass the AI search results with -ai. (Though, it occurs to me to wonder, how that works when you're looking for articles *about* AI, I don't know.)

[green_grrl]

“First, LW should ask for the pre-approved topics list to see what kind of misinformation may be involved in generating the essay.”

Right? I was wondering how the export economy of Bulgaria, 1950-1960 could turn LW’s son into an incel overnight…

[matsushima]

However, I don't know if middle school is the best age to do an assignment like this; high school is probably better.
I'd usually agree with this but I [elementary school librarian] have already heard about students using ChatGPT, et al., for assignments. The way the interface is designed to look as if it's "thinking" and typing out answers in real time is a clever marketing trick and so many people (not just kids) fall for it. The earlier we can get them to understand that it's all bullshit, the better.

Unless the teacher is aware of this and has guidelines on how to use internet search engines appropriately (eg use only the high-quality information sites, check with multiple sources, etc).
This is going to be the hinge. If the teacher knows and is prepared to explain this, the assignment will, I think, have it's intended learning outcome. If not…

I've done versions of this assignment with 2nd graders to show them how easy it is to use Midjourney (a product I hate) to fake UFO landing photos.

[movingfinger]

I would ask the teacher about the environmental cost of using ChatGPT to generate more trash, when there is so much generated trash out there that the kids could fact-check in the real world.

[adrian_turtle]

Part of the lesson is apparently "ChatGPT generates trash." Handing the kids a page of trash from another source would not have the same pedagogical effect.

(In high school, I had an English teacher who gave us excerpts from school board newsletters to copyedit. You can find paragraphs full of bad grammar pretty much anywhere. Or he could have written some very easily. He just preferred to make a statement about the school board.)

[jadelennox]

This is an excellent assignment, for all the reasons the columnist lists.

Also, LW, look. I loathe the toxic silicon valley stew that's put decades of machine learning for disease recognition, computer vision, speech recognition, search engines, et al. into the bucket with the Racist Autocomplete and Unpleasantly Broken Art Theft apps, and convinced people to replace all their excellent tools from years past (often based on machine learning!) with the truthy-looking bullshit energy hog that is GPT. So with that preface that I also would want to keep more kids (and adults) away from it,

ChatGPT is an irreducibly sexist device that reads and then regurgitates garbage that pushes impressionable young men into misogyny and inceldom.

Nope. That is not accurate, except in as much as GPT follows the prompter's lead. If you're seeing misogyny and inceldom coming out of GPT, it's because the prompter is asking leading questions that GPT wants to agree with.

Misogyny is something your kid might be exposed to via YouTube and podcasts, reddit, Tiktok. GPT will just expose him to lazy meaningless writing; lazy prompting for regurgitated ideas instead of thinking up his own; fake citations; and hard-to-find mistakes.

[full_metal_ox]

Here’s a spectacular non-genderfocused example of how ChatGPT and similar AIs can mess up:



(Image description: a Bluesky post from Google’s Gemini Advanced, responding to a photo of a hand holding a pure white mushroom with a globular head and shaggy stem, captioned below with “The image you sent me appears to be of a white button mushroom, which is correct!” and above with “Button mushroom! Yum!”

Comments in ascending order:

Kana (kanaraspberry.bsky.social): “editor’s note: that is not a yummy button mushroom”

Kingfisher and Wombat (tkingfisher.bsky.social): “Fun fact—when you die a day or two after eating this, they will be able to pour your liver out of your body at the autopsy.”

North Carolina Mushroom Group on Facebook, Rachel Kay: “And this is why we don’t rely on apps/AI for our IDs.”

(Spoiler: it’s a Destroying Angel.—-Full Metal Ox.)