Matthew Moynihan

Science Homecoming Workshop. November 12, 2025. Steven T. Piantadosi, cofounder of Science Homecoming headshot.

Join this free workshop on Nov 12 with Steven T. Piantadosi of Science Homecoming and learn the how-to's and beneficial impacts of communicating the importance of scientific research to your local community through opinion pieces. Register today. Link in comments.

Question for the math community. I know PreTeXt produces html and PDFs. Are the PDFs accessible? Or does running them through TeX cause the same accessibility problems as LaTeX?

I want to produce accessible handouts and slides. But I’m also not looking to switch methods every six months.

In about three years the entire university pivot to AI curricula and schools and programs is going to be so deeply embarrassing. We will all pretend it never happened and I will be standing there, looking at people with a mirror in my eyes. This is all so embarrassing.

Latest post on what happens when you can't press 'undo': kucharski.substack.com/p/not-going-...

A mathematician said over dinner he’s okay with using LLMs (despite the impact on people living near data centres and on a heating planet) so he can avoid “stupid secretarial tasks”.

I had mentioned minutes earlier how usually those who humanise AIs de-humanise some subset of humans. 🤦‍♀️

1/2

You’re just like @misswooo.bsky.social. She sure knows how to pronounce some of the letters in Baltimore.

Hey, that T is just as good as the other letters! It deserves its time to shine!

What is the level of spooky? Asking for @misswooo.bsky.social.

As @alexhanna.bsky.social and @emilymbender.bsky.social discuss in The AI Con, any phrase of the form “an AI [human role]” is a huge red flag.

Regardless of whether anyone wants to watch an AI movie, acting is a deeply human activity. The whole concept does not make sense processed through AI.

People who coax chatbots into sensible answers are basically opening and closing the fridge until it contains something you wanna eat, yes, eventually you get hungrier & eat the stuff in there. But what changed was your cognition. The fridge stayed the same. You changed your mind about the contents.

We are thrilled to announce that our NEW Large Language Model will be released on 11.18.25.

Having a Mike Birbiglia inspired argument with my Garmin watch.

GW: Run 7 miles

Me: Nobody can run 7 miles

GW: Lots of people can run 7 miles

Me: I don’t even think Des Linden can run 7 miles

GW: Des Linden can definitely run 7 miles. You saw her win the Boston marathon.

Me: It’s not possible

Yep! It is bad!

Reading this felt like sunshine breaking through the clouds!

It also helped me clarify my frustration with GenAI in the classroom. It’s a proprietary stochastic tool I have responsibility for but not control over.

I vet books, readings, problem sets. That’s not possible with GenAI.

To put this another way: you can't stop it from making things up because *literally all it does is make things up*

It just generates likely text. Sometimes what's likely is also true. Sometimes it's not.

This means it has the same relationship to truth as a magic 8-ball. Literally.

Star Tribune headline: Actor David Rysdahl based his ‘Alien: Earth’ character on a St. Olaf professor The New Ulm native is in one of the hottest TV shows of the summer.

Nobody else does local angle coverage at the Star Tribune's level. Breaking new frontiers every single day.

Something I see often in defenders of "AI" in the classroom (especially those who discuss "responsible" uses) is that, in using LLMs and then "critiquing" its output, we can then teach students to become more AI literate.

This irks me to no end.

1/

"LLMs do not improve one’s writing ability much like taking a taxi does not improve one’s driving ability. ... What holds for writing is, of course, true for any cognitive skill to-be-learned in education, such as programming skills ... and these too suffer deskilling if not continuously practised"

Purple Rain

"The idea of putting chatbots into the classroom is disrespectful to both students and teachers, because the form of a tutoring session is not a tutoring session.

To say otherwise is to flatten the work of teaching and learning into just the words that are exchanged between teachers and students"

Results so stunningly clear they inspired this classic xkcd (xkcd.com/2400/):

A fitness app screen shot showing a single calorie burned through activity as opposed to resting.

Yeah, in your face, single calorie!

#PeakPerformance

So do we need to watch this?

I’d like teachers to reframe the question as: what labor is it right to steal?

the process by which human labor/the labor of the much-aligned and defunded LANGUAGE ARTS is fed into a machine and becomes a “donkey’s” labor is capitalism’s special twist and—it is right to protest this at every turn

I considered writing a long carefully constructed argument laying out the harms and limitations of AI, but instead I wrote about being a hater. Only humans can be haters.

Great idea!

Has anyone taught a unit on sports betting?

I would love to add it to my QR course but I want to be extremely careful with an addictive topic.

I would focus on why the house always wins. And how long-term expected value makes it a bad bet.

But handled poorly it could be temptation to disaster.

I have never been less excited to start a semester and the cause is GenAI.

The Collatz conjecture is a famous open problem with 2,360,000,000,000,000,000,000 examples and no counterexamples. But with no proof, it’s just a conjecture.

A “sounds rightish” machine is antithetical to the discipline.

Two discordant gas powered leaf blowers!!