iiinova — where your digital self keeps evolving

Reposted by shiz

Joshua Bowman @thalesdisciple.bsky.social · 08/08/2026

Working on an A.I. policy for my calculus classes. Decided to be upfront about the current situation with LLMs and to give a heartfelt explanation for why the course still matters.

Al policy

Context

Artificial Intelligence, or "A.l.," Is a broad term used to refer to several different kinds of technologies. The kind most relevant to this class are Large Language Models (LLMs), which use mathematics to analyze large amounts of human-produced text and then generate new text that simulates what a human might write in response to prompts or questions. Some examples of LLMs are OpenAl's ChatGPT, Anthropic's Claude, and Google's Gemini.

Current LLMs can provide a solution to any calculus problem I give you. So why should we work on them, you might ask? Here's the thing: for the most part, the problems were already solved anyway. Calculus as a subject is nearly 400 years old. It has reached into virtually every modern scientific and economic field. And while I would like to believe that I have crafted a few new problems with some small amount of originality tailored to this class, for the most part I have curated a collection of exercises drawn from my years of teaching calculus. Those exercises come from textbooks, workshops, friends, and other fields of study, as well as my own invention.
And they nearly all have known solutions, often with fairly standard approaches.
Before A.I. became widely available, you could find those solutions in another textbook, or from a friend's notes, or through a repository on the internet.

The point of this class is not necessarily to solve problems that have never been solved before. It is for you to develop habits of thinking that are used by mathematicians, scientists, and policymakers-to develop a sort of mental toolbox that can be applied to new problems. If along the way you happen to find a new way of solving one of the class exercises, that's great! I am constantly learning new things from my students. But you will be bereft of discovery if you rely on a LLM for solutions to the calculus problems before you. Here I mean the discovery of having your own realizations, crystallizing your own internal understanding, claiming the knowledge in a way that cannot be taken away from you.
That ownership is what I want for you, and that is what the course is built around.

441923546

shiz @shirazkn.bsky.social · 20/07/2026

Remembering how 10 years ago I attended a math talk for fun, had not the slightest clue what the fuck he was talking about, pretended to take notes and drank chai in the back. Today I know enough math to follow along to pretty much any talk, at least for the first half or so

030

Reposted by shiz

James Webster @hcmjmagic.com · 17/07/2026

Constantly reminded that my brain is blindingly brilliant and beautiful because where normal people think "this sucks lmao" I think "wtf that is bad as hell"

5282

shiz @shirazkn.bsky.social · 17/07/2026

The sky is blue again! Never taking the blue sky for granted #wildfires

030

shiz @shirazkn.bsky.social · 28/06/2026

Anyone who thinks that AI solving Erdos problems signifies a permanent shift in what mathematicians will do or want to do is missing the point. Its like yeah AI will generate music and we’ll have AI jazz bars in the near future, but its not gonna stop musicians from wanting to write and play music

130

shiz @shirazkn.bsky.social · 28/06/2026

From Jim Simon’s’ wiki. Wonder if he knew about the sock market

010

shiz @shirazkn.bsky.social · 23/05/2026

Ugh Apple maps has ads now. Thats a slippery slope

120

shiz @shirazkn.bsky.social · 14/05/2026

Regrettably, I now have a twitter account x.com/shirazkn?s=21

x.com

Shiraz (@shirazkn) / X

110

Reposted by shiz

John C. Baez @johncarlosbaez.bsky.social · 07/05/2026

What's the most surprising fact you've learned in the last couple of weeks? I don't mind if it's quite technical. I just want to hear what you folks are being surprised by!

16295

Reposted by shiz

Carl T. Bergstrom @carlbergstrom.com · 03/05/2026

2. Whether or not this email was indeed AI-authored, it's worth taking a moment to reflect on why we write, where authenticity matters, and what happens when we use LLMs as communication surrogates to express our thoughts (or the thoughts think we are supposed to express) for us. From our course:

thebullshitmachines.com

Lesson 14: Authenticity

In this lesson we explore the concept of authenticity in human communication, and the ways in which LLMs get in the way.

611318

Reposted by shiz

ꭓle Ormsby 𓆙 @e-infinity.space · 30/04/2026

You may have heard that the Banach-Tarski paradox is a consequence of the group of spatial rotations SO(3) having a subgroup that is free on two generators. But what does that look like? Answer: Roughly like this! kyleormsby.github.io/F2-in-SO3/ 1/n 🧮

F₂ inside SO(3) ≅ ℝP³
Generators A = rotation by arccos(1/3) about x-axis, B = same about z-axis. Each reduced word w becomes a rotation in SO(3); we plot it as the point θn̂ in the ball of radius π, where θ ∈ [0, π] is the rotation angle and n̂ is its unit axis. Edges are true SO(3)-geodesics — some wrap through ∂B and reappear at the antipodal point (the ℝP³ identification).

Now transformed into the standard 'tree' Cayley graph of F_2.

39033

shiz @shirazkn.bsky.social · 02/05/2026

I've been working on the research side of this embodiment!! There's very few workplaces in the US where you gotta navigate a robot maze to get from your desk to the coffee machine

1231

Reposted by shiz

Brianna Lei @briannalei.bsky.social · 28/04/2026

The Kickstarter for The Crane Rider's Tale is now live! Fly passengers around a fantasy town inspired by Chinese folktales and unravel a mindbending mystery with multiple endings. Zero Escape meets Disco Elysium 🗡️🎲 LINK: www.kickstarter.com/projects/psy...

17739363

shiz @shirazkn.bsky.social · 08/04/2026

Project Hail Mary was unbelievabl y mid

120

Reposted by shiz

nat alison @tesseralis.site · 31/03/2026

Highly, *highly* recommend The Artisan of Glimmith (and not just because I made a few puzzles for it). I talk a lot about the beauty of logic/pencil puzzles and this is one of the most accessible ways to see it! store.steampowered.com/app/4160210/...

store.steampowered.com

Save 10% on The Artisan of Glimmith on Steam

The Artisan of Glimmith is a cozy, relaxing puzzle game about coloring stained glass and restoring the artworks of a kingdom above the clouds. Explore a world of handcrafted puzzles designed with inte...

0174

Reposted by shiz

jae holzman @jael.bsky.social · 07/03/2026

OpenAI head of robotics just resigned over company deal with the Pentagon saying… “Surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got”

Caitlin KALINOWSKI over X

I resigned from OpenAl. I care deeply about the Robotics team and the work we built together.
This wasn't an easy call. Al has an important role in national security. But surveillance of Americans without judicial oversight and lethal autonomy without human authorization are lines that deserved more deliberation than they got. This was about principle, not people. I have deep respect for Sam and the team, and I'm proud of what we built together.

13770842624

Reposted by shiz

lauren @lauren.rotatingsandwiches.com · 02/03/2026

mad at wire mother (dell work laptop). missing cloth mother (mac mini)

868928

Reposted by shiz

sawyer @sawyer-p.me · 22/02/2026

The long term economic bet seems to be keep general economic output relatively the same, automate with AI, replace workers and cut costs. How does that not just drive the price of everything to zero. The whole economy is people doing things for other people, I'm going insane.

132

Reposted by shiz

Robert Evans (the Only Robert Evans) @iwriteok.bsky.social · 22/02/2026

Just found out Fredrick Brennan died. I only met him in person once, but I always had a lot of respect for his personal growth and ownership of his past. More than anything, I respected his commitment to changing himself for the better. He was a good guy. So it goes.

371384111

shiz @shirazkn.bsky.social · 18/02/2026

Finally had a chance to play Ooo. The difficulty ramps up much slower than other puzzle games of this genre but this game is NOT fuckin around lol

140

Reposted by shiz

Chris Paxton @cpaxton.bsky.social · 17/02/2026

Building solar farms over bodies of water to cut evaporation and generate power

711119

Reposted by shiz

ꭓle Ormsby 𓆙 @e-infinity.space · 12/02/2026

Work tasks today included thinking about a magic trick and Catalan numbers at the same time — living the dream!

First page of 'Paperclip graphs', available at https://mathweb.ucsd.edu/~ronspubs/pre_paperclip.pdf.

Figure 1 depicts a magic trick that starts by "Putting two paperclips on a strip of paper shaped as an S curve."

Figure 4: From a strip with paperclips to sets of nonintersecting arcs.
(a) Strip with paperclips.
(b) Starting to pull apart.
(c) Making the strip taut.

Figure 5: Going from a set of arcs to a strip with paperclips.
Figure 6: Circular representation for arc configurations.

6458

Reposted by shiz

Chris Paxton @cpaxton.bsky.social · 07/02/2026

Businessmen, famous academics, politicians... open.substack.com/pub/persuasi...

0141

shiz @shirazkn.bsky.social · 27/01/2026

This is why I love math btw

010

Reposted by shiz

John C. Baez @johncarlosbaez.bsky.social · 26/01/2026

Not very urgent compared to some things, but okay: On MathOverflow I asked what has been done to make this question-answer website more welcoming to women: meta.mathoverflow.net/questions/63... Answer: something once, but nothing now. But some preferred to argue against the very question. (1/2)

meta.mathoverflow.net

What is being done to make MathOverflow more welcoming to women?

What if anything has been done, or is being done, to make MathOverflow more welcoming to women - or at least less unwelcoming to women?

1161

shiz @shirazkn.bsky.social · 23/12/2025

@skatestory.com is sooo fire I feel like I did the first time I played need for speed or burnout as a kid 🥹

010

shiz @shirazkn.bsky.social · 30/11/2025

Outlook for mac comes configured with a default "Get Outlook for Mac" signature.. I want 30 minutes in a room alone with whoever thought that that was a good idea

030

Reposted by shiz

Freya Holmér @freya.bsky.social · 26/11/2025

ahah IT WORKS look it's a C² dual quaternion B-spline!! this is interpolating *poses* in dual quaternion form, coupling position and rotation, instead of treating them independently (this is a really neat thing!!)

2355953

Reposted by shiz

Katie Mack @astrokatie.com · 21/11/2025

It’s widely known (and, I think, pretty uncontroversial) that learning requires effort — specifically, if you don’t have to work at getting the knowledge, it won’t stick. Even if an LLM could be trusted to give you correct information 100% of the time, it would be an inferior method of learning it.

8755851573

Reposted by shiz

Today in Simpsons Screenshots @todaysimpsons.bsky.social · 21/11/2025

~ November 21 ~ 2019: The Cybertruck is launched.

1144

shiz @shirazkn.bsky.social · 16/11/2025

When you try to place the flat torus in R^3, call that an embed fail

110

Reposted by shiz

Carl T. Bergstrom @carlbergstrom.com · 15/11/2025

Bear to Take Huge Dump in Forest

Justice Department to Investigate Epstein Ties, but Not to Trump

622218

shiz @shirazkn.bsky.social · 02/11/2025

Started a new blog post on Principal Bundles at www.shiraz-k.com! It's a long way away, but I'm excited about how this illustration turned out: #MathSky

A math diagram describing a "coset space", viewed as a fiber bundle

150

Reposted by shiz

Dr. Courtney Gibbons (she/her/xlookup) @courtneygibbons.phd · 29/10/2025

Relying on LLMs when we have specialized tools for tasks... it's so lazy! The best AI applications are specialized models (eg drug discovery machine learning, handwriting recognition vision models for sorting the mail) used for specific tasks. LLMs are not generalists!

151

Reposted by shiz

Socks the Macro Wolf Dad @socksthewolf.com · 27/10/2025

Python was awarded a funding grant, funding grant asked Python foundation to remove all diversity and inclusion initiatives they have. Python foundation said no and rejected the grant. If you use Python, send a few dollars to the charity to keep it going. I’m pushing a dono on behalf of ScamGuard.

2185854781

Reposted by shiz

Elke Stangl (elkement) @elkement.art · 27/10/2025

My version of this - I called it Dirac's Belt Swimming in Dirac's Sea :) Inspired by Roger Penrose's book The Road to Reality. Watercolor and acrylics #SciArt #MathArt #PhysicsArt Math brackground (plus early sketch of this) in this blog post of mine: elkement.art/2021/02/15/d...

Painting illustrating Dirac's Belt Trick: Undoing a 720° rotation of a belt fixed at one end with a kilogram weight. Four phases of the transition painted in different colors. By elkement 2024.

1252

shiz @shirazkn.bsky.social · 25/10/2025

#mathsky tip, if you're on MacOS consider getting Alfred (blazing fast drop-in replacement for Spotlight without AI bullshit) and set it up so that "w <search_term>" directly pulls up the Wikipedia page! Only downside is that you seem to have to spell the page exactly right. @alfredapp.bsky.social?

330

shiz @shirazkn.bsky.social · 24/10/2025

"Misuse as a Use of Language" from Jan 2023. Having fun reading my own writing, because back then I had soo much motivation to introduce the uninitiated reader to 20th century mathematics. This post leads into Gödel's incompleteness theorems.. shiraz-k.com/posts/langua...

shiraz-k.com

Misuse as a Use of Language

This is my first post! It discusses the question of whether spoken and written languages like English could be ’logical’ by design. I will break this post up into two parts. The first one does not req...

000

Reposted by shiz

Oisín McGann @oisinmcgann.bsky.social · 14/10/2025

‘96% of healthcare professionals said the model helps to reduce children’s anxiety, and 46% reported a lesser need to use sedation after the children played with the set.‘ ‘Lego doesn’t sell the sets and instead has donated more than 10,000 of the kits to hospitals around the world.’

5387161

shiz @shirazkn.bsky.social · 14/10/2025

Spent a lot of time today playing around with @diffgeom.com's incredible math visualizations www.diffgeom.org/software/ind... My favorite was the Torus Sketch! Makes me wanna learn Javascript

diffgeom.org

Andrew D. Hwang: Interactive Math Software

131

shiz @shirazkn.bsky.social · 13/10/2025

They call it a window because it keeps the wind owt

130

shiz @shirazkn.bsky.social · 13/10/2025

@skatestory.com is just incredible on all counts. Dude has been developing it for as long as EA has been developing their silly fortnite reskin

110

shiz @shirazkn.bsky.social · 10/10/2025

Asians abandoning East to follow degree in instrumentation (8) #cryptic #crossword

010

Reposted by shiz

Dave Richeson @divbyzero.bsky.social · 09/10/2025

New blog post! Through a sequence of images, I verify that the unknotting number of the connected sum of a (2,7) torus knot and its mirror is less than 6: I show that this first image is the connected sum, and after changing those crossings, it produces the unknot! divisbyzero.com/2025/10/08/t...

A drawing of the connected sum of a (2,7) torus knot and its mirror image (much modified).

An image of a crazy knot I claim is the unknot.

3246