Dr. Casey Fiesler

YouTube video by Oliver Richman Day 576 - Bunnies - Oliver Richman

He’s been writing a song a day for 500+ days. Gen AI could never be this good.

There were bunnies that were jumping on a trampoline
And I just learned that they weren’t real
If a bot can inhabit an unknowing rabbit
It might manufacture the way you make me feel

youtube.com/shorts/IJfCr...

Oh totally, one of the arguments I make in that blog post about blogging is that it allows you to accurately describe your paper before journalists and other folks on social media get to it :)

Oh I'd forgotten about that!! You're totally right, and I saw all the backlash about that at the time.

Amazing!

Why (and how) academics should blog their papers Encouragement and tips for writing accessible, public-facing versions of your research papers

✨ MAKE YOUR RESEARCH FINDINGS ACCESSIBLE ✨

I promise you don't have to make TikToks. Maybe a Bluesky thread or a LinkedIn post?

Or my favorite option... blog your papers!

cfiesler.medium.com/why-and-how-...

Lay audiences unable to access scientific literature want to learn from you. 🥰

This paper about science communication on TikTok found that videos summarizing research papers got the most engagement, "potentially reflecting high demand from lay audiences who are traditionally unable to access scientific literature." www.sciencedirect.com/science/arti...

To me this suggests...

Yep that sounds about right. You would probably find the conversation happening around this right now very interesting heh.

awesome, thanks!

Exactly the kind of thing I was looking for, thank you! (Also it should have occurred to me to just look at what else Cynthia had published. :) )

Thank you! I’m doing some work related to educational accusations of AI use so relevant to that as well.

/cc @jeffreybigham.com thought you might know or know someone who would :)

There is some serious Discourse right now on TikTok about AI + art + accessibility. A big creator made a video about AI being a way for people with disabilities to create art, and folks are really coming for him. I’m curious if there is HCI research on this topic especially re: attitudes!

Your glasses-forehead winkle combo means your face was destined to hit peak beauty at 80+ years old

I genuinely have no idea if this is supposed to be a compliment or an insult. 😂

I was trying to think of something a professor/researcher could use ChatGPT agent mode for, and the first thing that came to mind was finding paper reviewers. Scan paper. Search literature. Find email address. Log into manuscript system. Send invite. But without oversight I find this horrifying. :-\

YouTube video by Casey Fiesler Covert racial bias in large language models re: decision making. Paper published in 2024.

Response to this request for AI papers has been pretty quiet so far but also I figured I might as well provide an example of the kind of thing I mean re: TikTok style explainers. :) www.youtube.com/shorts/IFQeq...

I would love if you all would give me suggestions for papers about AI well suited for TikTok style explainers! Stuff that’s important or interesting or just good for more people to know about. (It’s ok if it’s your own paper!)

Well it’s definitely not cheating in my class because I allow them to use AI—for precisely these kinds of reasons. That’s why I’m thinking so hard about this.

That’s a quite different way than I think of “policing” but sure, that’s much more reasonable.

I’m reluctant to require AI use in an ethics class. :)

I really hope that people can take the time to find the amazing students who care about learning and are doing so!!

I think I am far more ethically obligated to make sure that students who deserve A’s get A’s than that students who don’t deserve A’s don’t get A’s.

Oooh I don’t like that those were the lessons of this. I think #2 especially is a really unfortunate takeaway.

I don't give exams. :)

Yeah that last bit is a complete non-starter. :-\

Well this is a technology ethics class, so we will be talking about those things at length.

The reason I don't want to be the AI police is because there is no reliable, accurate way to detect AI use.

Thanks for the thoughtful responses! I haven't decided yet about the book assignment referenced in this thread.

By the way, I do think it's a good/useful assignment (this issue aside) and get good feedback on it especially from PhD students. Feel free to use it yourself, works for any class!

I completely agree, which is why I noted not relying on assessment. It essentially becomes pass/fail, where the point isn't assessment, it's just for them to do stuff.

It does inherently create a stronger attendance component of the grade than I would typically have though.

Bottom Line: What Actually Trips Me Up Strategy Can I Fake It Easily? Why? Require personal experience or identity ❌ No I have no lived experience—any answer will be vague or hollow Ask for class-exclusive materials ⚠️ Sometimes I need to be spoon-fed the content to reference it accurately Require local/unpublished primary sources ❌ No I can’t access them and I won’t make up specific, verifiable data Close reading of rhetorical strategy ⚠️ Sometimes I try, but often give formulaic or shallow readings Identify subtle flaws or nuanced disagreement ❌ No Hard for me unless the critiques are already in the discourse Require multi-step, feedback-based processes ⚠️ Sometimes Possible, but hard to fake actual in-class interaction

I mean it's a decent analysis (though again, took me SEVERAL iterations of "are you sure about that?") but not actually that helpful for me in practice.

So yep overall I basically agree with this. bsky.app/profile/mkfe...

But also, I imagine it's demoralizing for those students who spend hours and hours working on something only to get a similar grade to students who use an LLM--even if they do appreciate the opportunity to learn.

I don't actually see why this is all that helpful since it's easy enough to just re-type things. (Plus this definitely falls into the "AI police" category--I don't actually want to ban use of AI.)