Remind me

[soc_puppet]

I have to rant about my philosophy teacher tomorrow. He seems like a genuinely good guy, I'm enjoying the class, but his opinions on ChatGPT are...

Look, he suggested it as a study aid for our first test, coming up next Thursday. Anyone know if that "Tell me how many times the letter 'X' appears in [word]" trick still works? I'm thinking of emailing him a suggestion of that, and also that he recommend all students who consult ChatGPT save or print off their results. My hope is that, if ChatGPT gives them bad info to study off of, having proof that their incorrect response is because of ChatGPT will result in fewer points lost for the students. (And hopefully, ultimately, the teacher understanding why I keep visibly and emphatically shaking my head whenever he suggests relying on it as a study aid in class, but I'm not holding my breath.)

...Actually, I think that's about all the rant I need right there. Describes my problem succinctly (aside from how he might be a little confused about what "passing the Turing Test" actually means WRT machine intelligence or lack thereof), gets to what I think might be actionable and a possible solution, etc. I might talk about the fun stuff from his class tomorrow, instead.

Comments

Well ...

[ysabetwordsmith]

He probably thinks of it as a glorified flashcard set, because it is is pretty good at call-and-response routines. The problem is, as you pointed out, that AI is highly prone to factual errors.

Conversely, if students team up to study together, then they both learn from asking, answering, and discussing study questions.

Re: Well ...

[pikkugen]

This. It's better to study with actually intelligent company than a stack of hallucinating cards.

Re: Well ...

[soc_puppet]

It's actually for essays (our teacher showed us the two question options early, and encouraged anyone who was nervous to plug them into ChatGPT to get a better idea for how they'd want to format their essay, which. Doesn't still doesn't sound great to me), but yeah, the hallucination thing is a major worry for me.

Re: Well ...

[soc_puppet]

I probably misrepresented this a bit; he suggested it as a study aid for an essay test. He gave us the essay questions in advance, and suggested that, if we were nervous, we could plug them into ChatGPT and see what it spit out to use as kinda like. A guideline? Inspiration? Something like that.

I still feel like he would be better served by getting permission from prior students to use old essays, or by designing an example himself. But then, I worry about things like ChatGPT suggesting that Plato came to break Socrates out of prison, not Crito, or that Descartes came up with his refutation for global skepticism while on an astronomy related trip with Galileo, and students using those for their essays. After all, while the teacher has stressed that students should not use ChatGPT for their actual classwork, he has heavily implied that it's trustworthy!

Re: Well ...

[ysabetwordsmith]

In addition to your well-founded suspicion of inaccuracy ...

So while every other adult is screaming bloody murder about how using AI for schoolwork is cheating, this ignoramus is telling students to use AI. And then people wonder why students think cheating is "when the software says so."

Re: Well ...

[soc_puppet]

And can we talk about the irony of a philosophy teacher encouraging his students to outsource their thinking? Like. What the fuck. You don't become a good philosopher by letting other people, or machines, think for you! Even if the machines were actually intelligent, instead of just predictive text writ large!

Anyway. Possible sub-goal this semester of toppling his idea that AI is actually intelligent or useful. Primary goal is still to pass the class.

Re: Well ...

[ysabetwordsmith]

I have seen someone posting scripts where they asked a chatbot for Socratic scripts about ethics.

... wait, that was another teacher. 0_o

You should definitely stab that idea in the reality tunnel as much as possible.

[alexseanchai]

seen yesterday on Tumblr but fuck if I recall who reblogged it: Google search AI providing a series of differently wrong answers to the question "is it 2025"

[librarygeek]

It certainly feels like 2020 all over again, not some new year 2025! 😢🤷‍♀️

[soc_puppet]

I dunno, I'm seeing slightly fewer masked faces these days than I did at this same time of year in 2020. (Not nearly as many as I'd like, but here we are...)

[soc_puppet]

Excellent, another item for my toolbox 👍 Thanks!

[gwydion]

O.o

[soc_puppet]

Yeah, it's. I mean. At least he doesn't want anyone to use it for actual classwork, he just suggested nervous students use it to get an idea for how they might want to format the response to a long essay question. And has more or less implied that it's a reliable source, which. At least if he were promoting Wikipedia, it wouldn't be wasting so many resources!

[gwydion]

Ugh.

[jesse_the_k]

Wow

Any idea how long this person has been teaching? Cause I’m super horrified (and also haven’t been near a classroom since 1977, though there are many teachers on my d-roll)