pulling tarot cards until I get the answer I want

The telling detail rn is that AI users are much less interested in debating Doctorow about efficacy than in reading posts by Ronacher and Hughes about how to handle transformative change

new blog post! why do LLMs freak out over the seahorse emoji? i put llama-3.3-70b through its paces with the logit lens to find out, and explain what the logit lens (everyone's favorite underrated interpretability tool) is in the process.

link in reply!

a very large percentage of US doctors are using LLMs all the time, esp for quick research

because it’s effective and efficient to do so

Reminder. If a recession causes unemployment it will not be a reliable indicator of the effect of AI.

sweet time for sure. will ship papertrail next time I've got a day to hack around.

lots of interesting / strong ideas, like this one: Consciousness as Error Correction

x.com/Tetraslam/st...

then a multi-agent system works in an open-ended loop to refine the idea. the main agent can search w exa, call an adversarial paper reviewer to increase plausibility, or call a poetic inspiration agent to increase novelty. should probably add a terminal here 😈

Claude 4 consumes the lineage, then predicts what paper will come next. For the topic of promptable / reasoning segmentation models, Claude repeatedly predicted a cross-domain, universal encoder over many input types (lidar, satellite, RGB etc)

built a paper predictor at the exa / @anthropic.com novelty hackathon. uses 3M embedded abstracts and clustering to build a semantic lineage + trajectory for any idea. here I paste in a 2023 prompted segmentation paper, LISA, and progress through LISA++ and to LISAT (May 2025)

Datadog released an open-weight, state-of-the-art time series foundation model called Toto, designed for multivariate time series forecasting with an emphasis on observability metrics.

They also introduced BOOM, their time series observability benchmark.

I think the twitter devs vibecoded their infra

shoulda bought more puts

the limits of my language mean the limits of my world

fullstack developers be like

I was… probably not gonna last much longer at NPM. I was doing really poorly just trying to do what was apparently needed. I wasn’t on a PIP but I was very close to it.

And then I was just… given free rein and that’s when cacache/pacote/NPM5 & 6 happened

It was kind of incredible

if you’re a webmd regular, you’re gonna love deep research

Screenshot of top half of first page of paper. The paper is titled: "When People are Floods: Analyzing Dehumanizing Metaphors in Immigration Discourse with Large Language Models". The authors are Julia Mendelsohn (University of Chicago) and Ceren Budak (University of Michigan). The top right corner contains a visual showing the sentence "They want immigrants to pour into and infest this country". The caption says: Figure 1: Dehumanizing sentence likening immigrants to the source domain concepts of Water and Vermin via the words "pour" and "infest". The abstract text on the left reads: Metaphor, discussing one concept in terms of another, is abundant in politics and can shape how people understand important issues. We develop a computational approach to measure metaphorical language, focusing on immigration discourse on social media. Grounded in qualitative social science research, we identify seven concepts evoked in immigration discourse (e.g. "water" or "vermin"). We propose and evaluate a novel technique that leverages both word-level and document-level signals to measure metaphor with respect to these concepts. We then study the relationship between metaphor, political ideology, and user engagement in 400K US tweets about immigration. While conservatives tend to use dehumanizing metaphors more than liberals, this effect varies widely across concepts. Moreover, creature-related metaphor is associated with more retweets, especially for liberal authors. Our work highlights the potential for computational methods to complement qualitative approaches in understanding subtle and implicit language in political discourse.

New preprint!
Metaphors shape how people understand politics, but measuring them (& their real-world effects) is hard.

We develop a new method to measure metaphor & use it to study dehumanizing metaphor in 400K immigration tweets Link: bit.ly/4i3PGm3

#NLP #NLProc #polisky #polcom #compsocialsci
🐦🐦

aye bro what are yall doin

ok i pull up

yeah

lfg

800-588-2300 Gulf Coast

Frito Bay

McGulfin

cheesy gordita gulfwrap supreme

when my camera is off during the zoom it’s because im doing this