Matthew Moynihan
patternavoider.bsky.social
Matthew Moynihan
@patternavoider.bsky.social
Massachusetts Mathematician from Minnesota
“You’re now using Gemini on your iOS device”

The fuck I am!
a cartoon of batman with the words hmm weird nobody asked above him
Alt: a cartoon of batman with the words hmm weird nobody asked above him
media.tenor.com
November 20, 2025 at 7:25 PM
This is why I’ve bristled so hard at the push to incorporate GenAI in the classroom.

Me: Ok, how?

Them: You have to figure that part out too! And we’ll take the credit if you do.
A comment on the utterings of the techbros that applies equally as well to their products. They put the onus on the consumer to do the work to understand them and make their half-formed products useable. All while seeking to suck up all of our resources.
Why is the onus on *us* to interpret their babble as something clever and good rather than the onus being on *them* to make sense?
November 19, 2025 at 1:18 PM
I’m loving these Thanksgiving-themed episodes of cooking shows. Just a ton of “let’s add some heavy cream to that butter!”
November 16, 2025 at 3:35 PM
Love seeing this 15 months with an expired contract and learning back pay probably won’t come until spring.

I’m financially stable. But administration having an “aww shucks, this is how it goes” attitude while asking adjunct faculty to work for three semesters without any COLA is infuriating.
November 11, 2025 at 4:21 PM
Reposted by Matthew Moynihan
To state the obvious, doctors lawyers and priests are not products offered by a company that generates statistically likely text
November 10, 2025 at 3:57 PM
If capitalism is so great, why can’t I pay a reasonable amount of money to watch a single NFL game?!?!

Why isn’t the free market solving this problem?!?!
November 9, 2025 at 7:18 PM
Reposted by Matthew Moynihan
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.
November 3, 2025 at 5:00 PM
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.
October 22, 2025 at 6:01 PM
Reposted by Matthew Moynihan
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.
October 17, 2025 at 12:16 PM
Reposted by Matthew Moynihan
Latest post on what happens when you can't press 'undo': kucharski.substack.com/p/not-going-...
October 15, 2025 at 11:06 AM
Reposted by Matthew Moynihan
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
October 13, 2025 at 8:52 AM
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.
October 2, 2025 at 2:23 PM
Reposted by Matthew Moynihan
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.
October 1, 2025 at 9:55 PM
Reposted by Matthew Moynihan
We are thrilled to announce that our NEW Large Language Model will be released on 11.18.25.
October 1, 2025 at 2:38 PM
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
October 1, 2025 at 3:22 PM
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.
Finally! 🤩 Our position piece: Against the Uncritical Adoption of 'AI' Technologies in Academia:
doi.org/10.5281/zeno...

We unpick the tech industry’s marketing, hype, & harm; and we argue for safeguarding higher education, critical
thinking, expertise, academic freedom, & scientific integrity.
1/n
September 9, 2025 at 12:55 AM
Reposted by Matthew Moynihan
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.
How many times, in how many contexts, from how many internal and external researchers, or from how many CEO's are people going to have to receive this message before they believe it:

"Hallucinations" are an inherent part of the large language model architecture.
www.nature.com/articles/d41...
Can researchers stop AI making up citations?
OpenAI’s GPT-5 hallucinates less than previous models do, but cutting hallucination completely might prove impossible.
www.nature.com
September 8, 2025 at 6:38 PM
Reposted by Matthew Moynihan
Nobody else does local angle coverage at the Star Tribune's level. Breaking new frontiers every single day.
August 27, 2025 at 12:25 PM
Reposted by Matthew Moynihan
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/
September 8, 2025 at 3:41 PM
Reposted by Matthew Moynihan
"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"
September 6, 2025 at 1:52 PM
Reposted by Matthew Moynihan
September 2, 2025 at 9:04 PM
Reposted by Matthew Moynihan
"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"
It has been really interesting to attend UNESCO's Digital Learning Week (though unfortunately I'm not able to stick around). My public lecture from yesterday can be found here:

www.youtube.com/live/l-OWi6V...

>>
September 3, 2025 at 2:53 PM
Reposted by Matthew Moynihan
Results so stunningly clear they inspired this classic xkcd (xkcd.com/2400/):
September 1, 2025 at 7:32 PM
Yeah, in your face, single calorie!

#PeakPerformance
September 1, 2025 at 6:40 PM
Reposted by Matthew Moynihan
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
teacher on reddit saying she uses chat gpt "for the donkey work" and so it's not affecting her. feels emblematic: using it to do a class of work that's "beneath" one, reifying the idea that some work is beneath one, disconnecting process work from outcomes, relegating some work to "donkeys"
August 29, 2025 at 11:55 AM