Joshua Grochow

Here's the reality this example illustrates:

It's not even just about people blindly trusting what ChatGPT tells them. LLMs are poisoning the entire information ecosystem. You can't even necessarily trust that the citations in a published paper are real (or a search engine's descriptions of them).

Clopen

When you fold your pizza it can even be a closed-face sandwich

The current 27 quote posts on bsky.app/profile/oliv... are a pretty good summary of my thoughts on it so far.

Figure 6. The quotientope lattice for n = 4: all quotientopes ordered by inclusion (which corresponds to refinement of the lattice congruences). We only consider lattice congruences whose fan is essential. We have highlighted the cube (green), Loday’s associahedron [Lod04] (blue), another one of Hohlweg–Lange’s associahedra [HL07] (purple), the diagonal rectangulation polytope [LR12] (orange), and the permutahedron (red). Adapted from [PS19, Fig. 9].

This diagram should win some kind of prize for mathematical illustration. From arxiv.org/abs/2305.08471. 🧮

But was it made in TikZ? (/s /g)

In & near my area of expertise, I see a pattern that some experts save themselves time using GenAI, and more & more novices get led astray by it (and then waste experts' time finding & fixing their mistakes).

I take this as yet another sign that neither experts nor novices should be using it.

y'all I was approved for extra time accommodations today and it was super easy! I just lived with undiagnosed ADHD for 30 years, dropped out of school twice, suffered a lot of academic trauma, spent untold amounts on mental health treatment, and then took 16 months to fill out the paperwork!

I think the source code per se isn't the issue. I think part of the reason that Google Scholar is so (relatively) good is that they are using Google's index of the entire web as their first source of data. So they get to take advantage of the entire infrastructure of crawling, storage, search, etc.

Sex, lies, and iron deficiency: a call to change ferritin reference ranges Abstract. Iron deficiency is a very common and treatable disorder. Of all the tests available to diagnose iron deficiency, the serum ferritin is the most a

lol WHAT there’s no physiological reason women’s iron store range would be healthy at a lower level than men’s but they set the ranges based on populational distribution WITHOUT CONSIDERING more than 2.5% of women might be deficient!!

ashpublications.org/hematology/a...

Is there any realistic alternative because the profiles with pdfs and whatever are so useful? Shall we build one?

Remember folks, during winter every ten degree temperature drop lowers your tire’s PSI by 1 so it’s very important to go to the gas station and notice that the air thing doesn’t work and then do that five more times at five different gas stations

After staring at this plot for too long, I have convinced myself that

log(x) = lim_{m→0} (xᵐ-1)/m

is not such a bad definition of the natural logarithm. www.desmos.com/calculator/b...

#Mathsky

Are we aware that money is being spent to outsource mathematical inspiration to a plagiarizing water wasting machine? Instead of giving money and water to nurture mathematics in existing communities? This is a colonizer mindset. Life has to be more important than "discovery."

Started a thread in the other place and bringing it over here - I really think we should be more vocal about the opportunities that lay at the intersection of these two options!

So I'm starting a live thread of new roles as I become aware of them - feel free to add / extend / share :

The U.S. Is Funding Fewer Grants in Every Area of Science and Medicine A quiet policy change means the government is making fewer bets on long-term science.

The U.S. Is Funding Fewer Grants in Every Area of Science and Medicine www.nytimes.com/interactive/...

“Thank you for your effort, and for keeping us safe with your work.” She smiled SO genuinely, probably the most genuine smile I’ve seen in a while. How often we fail to thank the people who quietly keep our surroundings running well. It also smelled like lavender. We should thank more often

The anxious feeling of pushing code to production is so very similar to the feeling of pushing ‘submit’ for sending a manuscript to a journal 🫣 Even after 20 years as an academic

Am I the only one that thinks of the Death Star when hearing the phrase "fully operational"?

Once you gave it the statement am I reading the transcript correctly that the model output a correct proof independently?

Concrete, Pictorial, Abstract: Using the CPA Method in Upper Elementary Math Teachers can help students build deeper math understanding and confidence by introducing concepts in distinct stages.

Studies have shown multiple benefits of the CPA method (Concrete, Pictorial, Abstract): better test scores, deeper conceptual understanding, and boosts in mathematical confidence. 🌟

Here’s how to use it in your classroom.

#MathSky #ITeachMath #ElementaryTeacher

On the Irreducible Character Degrees of Symmetric Groups and their Multiplicities We consider problems concerning the largest degrees of irreducible characters of symmetric groups, and the multiplicities of character degrees of symmetric groups. Using evidence from computer experiments, we posit several new conjectures or extensions of previous conjectures, and prove a number of results. One of these is that, if n >= 21, then there are at least eight irreducible characters of S_n, all of which have the same degree, and which have irreducible restriction to A_n. We explore similar questions about unipotent degrees of GL_n(q). We also make some remarks about how the experiments here shed light on posited algorithms for finding the largest irreducible character degree of S_n.

"On the Irreducible Character Degrees of Symmetric Groups and their Multiplicities" by David A. Craven. #ExperimentalMath #RepresentationTheory #SymmetricGroups #MathSky

I love it when people talk to me as if I neither am disabled nor actually study this in my own work. But let me pitch someone else's work here, Doron Dorfman's "Fear of the Disability Con."
Sharing it for everyone feeding the fear of the disability con right now:

papers.ssrn.com/sol3/papers....

Adding "-ai" to the end of my searches becoming second nature on normal google now they're adding it to scholar too 😭

AI is Destroying the University and Learning Itself Students use AI to write papers, professors use AI to grade them, degrees become meaningless, and tech companies make fortunes. Welcome to the death of higher education.

„CSU—US’s largest public uni system—went all-in with a $17mill partnership with OpenAI. . .CSU unveiled its grand technological gesture just as it proposed slashing $375mill from its budget. While admin cut ribbons on AI, they were cutting faculty positions, academic programs, student services.“

Dear [TEACHER], I want to make sure we're on the same page for the rest of the year, because I was surprised and disappointed to hear that you uploaded student projects to ChatGPT. We have always found [SCHOOL] to be a place that encourages students to express themselves creatively, to be responsible and aware of how their actions affect others, and to be intellectually honest – doing their own work, rather than copying from others. I know ChatGPT has been heavily marketed as an educational tool, but everything about it is the opposite of those values. AI tools are built by training models on stolen work. They have huge and terrible environmental costs, because their data centers generate enormous amounts of pollution and are cooled with drinking water. They're also incredibly dangerous: tools like ChatGPT have taught children how to start fires and even encouraged them to commit suicide. And in a classroom setting, it teaches students that a machine can express their ideas better than they can. I told [ CHILD ] I was interested in seeing what they had to draw, and my heart sank when they told me "but I can't draw that pretty." Children deserve to draw what they imagine. These environmental, ethical, and pedagogical concerns are why I don't want [ CHILD ] to use AI tools (or for you to use them on their behalf). For the same reasons, I respectfully ask that you stop using them in your classroom. If you do plan to use AI this year, please send home a permission slip requesting consent. There are important privacy concerns when you upload student work into a for-profit service like ChatGPT. Thank you, [ PARENT ]

I ended up drafting an email that didn't really lean on research specific to grade school education, but expressed my concerns in the most basic terms.

I removed the personal details and have shared it as a template here for folks who might find it useful:
bit.ly/TeacherEmail...

Google Maps has added AI information to its search results and there's no way to turn it off.
#Enshittificatiom

I cannot stress enough just how much wide LLM adoption is harming the human capacity to *try*. To just do things on your own, to think critically about a topic or a problem, to read and synthesize ideas. These are such important skills and we’re just giving them up.

Why would we want to accept something that removes human agency and critical thought? Like, quite literally, we are building systems that are making us less capable and less interesting human beings.

‘Reverse Mathematics’ Illuminates Why Hard Problems Are Hard | Quanta Magazine Researchers have used metamathematical techniques to show that certain theorems that look superficially distinct are in fact logically equivalent.

What does counting pigeons have to do with Turing machines? More than you might think. In @quantamagazine.bsky.social, my latest foray into the wild world of meta-complexity, this time through the lens of mathematical logic: