Naomi Saphra

I am sorry to drop this bomb when it’s almost sold out but the Boston Ballet‘s Winter Experience is mind blowing and I just got tickets to see it a second time.

Yeah I think this would not actually help enough. Explaining site navigation is something they are still unbelievably bad at.

Every time I have to follow the instructions of an AI "help" search result on any app or website I know I have no chance. I have not seen a help menu that actually points to an existing, real menu item in at least a year.

Some people showcase intelligence not by discussion but by playing a game called "find the flaw". It is a very easy game but irresistible to junior paper reviewers, critical theorists, and terminal posters. If you know who's playing, it explains everything from reviewer 2 to "you hate waffles".

I don't know how to explain this but Casablanca is a movie for adults.

There are lots of movies built on an element of fantasy about being young or brave or defying the odds.

Casablanca is not that. Everyone in that movie has back pain and they have all just accepted it.

📣 Excited to announce the 2nd edition of our workshop
“Agent-Based Models in Neuroscience: Theory, Autonomy, Embodiment & Environment”
at @cosynemeeting.bsky.social #CoSyNe2026!!
🧠🤖🌍🪰🐟🐭💪🧘🏃

🗓️ March 17, 2026
📍 Cascais, Portugal
🔗 Speaker lineup and schedule: neuro-agent-models.github.io

I check if they cited me. If not, I know I would have done a better job because I would have cited me

I got a copy of this from Red Emma’s Baltimore a decade ago and it‘s sat on my coffee table since.

I knew working on suicide research was going to be heavy.

But what's actually starting to get to me is reading all of these chatbot suicide laws and legislative proposals that call for measures that have been shown to exacerbate crisis.

www.governor.ny.gov/sites/defaul...

When I get bored with science, it's because I've read the same paper three times from different authors whose media diet consists of corporate PR releases and unhinged slop summaries for viral arxiv drops.

I also wasn't raised with videogames, so I find media about them very boring tbh. I am planning to book club Seven Games with my gf soon (BOOK THREAD PREVIEW???) but it doesn't focus on videogames.

10.5 (beautiful, floofy) and 8lb (beautiful, sleek)

I can't play video games because my hands don't work

Gotta get in front of this scandal: I did an interview with @shwartzzzivravid.bsky.social where I said isometric instead of isotropic like an idiot. I am a very simple language model.

I deleted social media from my phone and blocked my browser so now I keep refreshing my Litter Robot (tm) app to check how much my cats weigh.

When shaping your research agenda, your objective is to find the weirdest niche possible that still has the potential to change everything.

SIGBOVIK has a Bluesky now! Follow to learn more cutting-edge research from the world’s most comedic and occasionally scientific academic conference

Donald Knuth asking Claude to update plan.md feels like some kind of time warp

if you know what's in the footnote, it's time to schedule your colonoscopy

My issue with every representation alignment paper is that you only believe A+C and B+D are more similar than A+B and C+D if you're already humanpilled by training on multimodal language tasks like image classification (YES, IT IS) or captioning. What's important to humans is not visually obvious.

Sorry, I think this is you reading a bunch of actual slop posts. Not everyone converging to it in their writing.

Title, author list, and two figures from the paper. Title: The Aftermath of DrawEduMath: Vision Language Models Underperform with Struggling Students and Misdiagnose Errors Authors: Li Lucy, Albert Zhang, Nathan Anderson, Ryan Knight, Kyle Lo Figure 1: On the left is a math problem, where students are asked to draw x < 5/2 on a number line. The right side shows two example student responses that differ in correctness. DrawEduMath pairs each math problem with one student response, and prompts VLMs to answer questions about the student response. Figure 2: VLMs consistently perform worse on answering DrawEduMath benchmark questions pertaining to erroneous student responses. Performance on non-erroneous student responses is labeled with specific VLMs’ names; that same model’s performance on erroneous student responses is directly below.

Models are now expert math solvers, and so AI for math education is receiving increasing attention.
Our new preprint evaluates 11 VLMs on our QA benchmark, DrawEduMath. We highlight a startling gap: models perform less well on inputs from K-12 students who need more help. 🧵

I do watch some anime, but I don't usually watch shonen anime like AOT---except, weirdly, sports anime. But the best sports anime is actually a shoujo romance: Chihayafuru, which is my recommendation back!

Oh this is the phase transition that shows up a lot in Lenka Zdeborova’s work! (She’s giving a talk this Friday at 230pm at Harvard SEC, it’s just a 40-50 min walk from BU.)

I haven't seen Pluribus but I LOVE hiveminds, I've just been a lil too stressed to start an intense show.

Didn't mean to full on resurrect this thread, but have you read Leech (Hiron Ennes)? It's narrated by a parasitic hivemind who controls every doctor in the world.

I'm actually reading A Distant Mirror now, it is pretty great so far!

Senators Reveal That Feds Have Secretly Reinterpreted The PATRIOT Act We had mentioned, briefly, the amendment that Senators Ron Wyden and Mark Udall had tried to push for with relation to the renewal of controversial provisions of the PATRIOT Act, however didn&#8217…

The deal Sam signed is the kind of deal someone who doesn't know how the NSA lies by telling you what you want to hear, but then secretly changing their definition of the plain English words in the contract.

www.techdirt.com/2011/05/26/s...