Well, I stumbled on this gem this morning. #jazz #blues
OpenAI has figured out that people prefer to hear what they want to be true over what is actually true. They prefer insincere compliments over honest feedback and constructive criticism (so long as there is no way to find out that the comments are insincere).
I don’t think that technology can solve everything - even AI (LLMs). This is an incredibly important talk and discussion about how chatbots are changing social interactions, including finding friendship and falling in love with them. What principles should be adopt to avoid dystopia?
My message to the graduating class of 2025
Friday, May 23, 2025
I don’t usually give a talk to my seniors because I don’t feel like I have anything insightful to share. But the other day I woke up at 4:45am with an idea. I threw the the slides together before work and was only able to practice part of it before I gave it. Here it is, in its entirety.
The math classroom that we math teachers enjoy being part of won’t be optimal for mathematical novices. We should generally resist the urge to use that environment as our aim.
(This remains true until and unless students have mastered the concepts that they will be working with.)
Math in Action 2025
Friday, February 21, 2025
I have the privilege of sharing a wide range of ways that I use Desmos, especially the Desmos activity builder, at the Math in Action conference on Grand Valley State University’s campus! Whether a teacher is brand new or has been in the classroom to decades, my hope is that all participants can find something valuable and usable in their classroom! Here are my “slides,” which is built in the Desmos Activity Builder.
Everyone is probably already aware of this, but it was only a matter of time before advertising came to AI chatbots. Perplexity seems aware of the problems with this, and is going to try to mitigate it. We’ll see. I suspect the game theory will drive a race to the bottom similar to social media.
Although, powerful open source models may be disruptive enough that advertising in AI LLMs doesn’t have the influence like it does in social media. Imagine if nostr and twitter started at the same time and nostr didn’t have to fight the network effect of most people having their social graph in FB, twitter, IG, etc.
To fully deliver on our mission to spark the world’s curiosity, we need to invest in building not just a beloved product, but a robust and self-sustaining business. That’s why starting this week, we will begin experimenting with ads on Perplexity.
Before getting into the details, we want to highlight our guiding principle: the content of the answers you receive on Perplexity will not be influenced by advertisers. Users come to Perplexity for a more efficient, uncluttered, and unbiased search experience, and that isn’t changing.
Ads will appear in the US to start and will be formatted as sponsored follow-up questions and paid media positioned to the side of an answer. Here’s an example of how they will appear:
Good stuff.
Prioritize:
- Sleep
- Learning
- In-person interaction
- Exercise
From the book “math without numbers” by Milo Beckman
Interesting take on AI hype from Freddie deBoer.
Some say AI is the greatest invention of all time. I don’t get it.
Anyone who cites AI as the pinnacle of human ingenuity, above plumbing, should try spending a month without the latter.
I exercised for 300 days in a row - and then I crushed my ankle
Monday, January 27, 2025
Around spring break of last year (2024) I decided to exercise every day. I worked out fairly consistently, 3-4 days a week, for close to two years at that point, but I decided to commit to every day for a few reasons. First, I learned that natural muscle loss from aging can start in your mid thirties. Despite my insistence that the final season of The Office ended just a few years ago (it was 2013), by most chronological accounting methods I’m 35 years old.
Quote from Seneca that resonated with me.
“I say, let no one rob me of a single day who isn’t going to make a full return on the loss.” —SENECA
It’s not just red dye no. 3. It’s all our stuff.
RFK Jr. has the potential to do what Sinclair did in the early twentieth century: to shift the vibe, and spark the public into demanding companies remake their products for the twenty-first century. America cannot claim to be exceptional while making inferior, toxic products. The process will take decades to fully mature, but Americans have never been afraid to insist on better. It’s time that we do that now, and demand a Great Reformation.
Hopefully TikTok will be banned as scheduled.
And even if parental controls worked and parents chose to shield their kids from bad stuff, they can’t because TikTok’s content moderation is poor. An internal study found that the “leakage rate” (of bad stuff getting past moderators) is as follows: 35.71% of “Normalization of Pedophilia” content; 33.33% of “Minor Sexual Solicitation” content; 39.13% of “Minor Physical Abuse” content; 30.36% of “leading minors off platform”; 50% of “Glorification of Minor Sexual Assault”; and 100% of “Fetishizing Minors.” For those who think that social media is relatively harmless, we urge you to read the quotations and internal studies described below, in which employees of TikTok discuss the vast and varied harms that they are causing to literally millions of American children each year.
Easily use AI to convert handwritten notes to text
Wednesday, January 1, 2025
It dawned on me today as I was taking notes on a video about weekly reviews that I could probably upload a picture of my notes to Claude and ask it to transcribe them into text. And, even with my poor handwriting, it did a great job. I also asked it to format it as a bulleted list with sub bullets, which it did. If my handwriting was slightly better it would’ve done an even better job.
Creating a Critical Thinking Curriculum
Tuesday, December 31, 2024
A couple of years ago I was on Drew Perkins’s podcast to discuss a curriculum that Will Reusch and I designed called Critical Thinking Across the Disciplines. I realized that I never actually posted about it on my blog, so here it is. Here’s the podcast description: Drew Perkins talks with Will Reusch and Zach Cresswell about the Critical Thinking Across the Disciplines curriculum they’ve created and are piloting with funding through Heterodox Academy.
Testing youtube shortcode. Here is the video from my previous post.
I thought that large language models might asymptote in their capabilities. If that’s true then it doesn’t seem like we’ve reached that limit yet.
AGI or not, we can't close our eyes
Monday, December 30, 2024
There’s been a conversation about when AGI will arrive for a long time. That time horizon has shrunk since chatGPT came on the scene, obviously. However, for a while I thought maybe these models would asymptote. I’m not convinced of that anymore. I think AGI, or something close enough, is hurdling towards us. We need to start grappling with these questions (and by “we” I mean the general public, as I’m sure many in the industry have these conversations frequently):
An incredibly snowy and peaceful Christmas Eve.
