Someone in Nature finally noticed how flawed this paper was. It only took them 1 year. www.nature.com/articles/s41... When it came out, I found multiple inconsistencies in their data just by looking at it for 20 minutes 👇

Personal news! I am retiring from UC Irvine at the end of June 😎 Retiring from UCI, but not retiring in general. What comes next is a combination of independent consulting 🔆📊, running my winery Quinta das Queimas 🍇🍷, and doing research at INESC TEC 👩‍💻. I will still be affiliated with UCI 🎓 (Emerita).

Creating a quiz for my course, I’m very tempted to include questions with “wrong answers only”

GenAI is not going to go away. It's super powerful tech; it's only going to get better. There will be adjustments -- e.g. cleaning up the legal mess -- but there is no going back. Its use will move from the engineering and business realms to the psychology&cognition, policy, and political realms.

It's quite ironic that the country that doesn't have national identity cards (because they are considered too intrusive & controlling) ends up with a private company doing that and much, much worse!

Wow

I really have no idea how to advise CS students these days on matters of career. The job market is brutal for juniors. Almost feel like telling them that if their goal is to get a job, they should switch majors or quit College and enroll in a trade school.

Good riddance, Hungary!

I always wanted to be an astronaut ever since I was a little girl, but clearly that's not gonna happen, and that makes me sad. I'm now considering the closest experience: 🍄 www.nasa.gov/wp-content/u...

As a research project, I built a needed tool with Claude Code. I though it would be a disaster, but It wasn't. I have some complicated feelings about it.

Rubrics these days... But, you know, real learning has always been about asking the right questions. It's harder to teach that, though, that's why it's typically done at graduate level.

🍓🐶🍓

Much will be written in years to come about the tech world's gradual shift from benign progressive irreverence to plain evil. Or maybe all has already been written since Macbeth, and we're just stuck in this human condition forever and ever.

Wow, what an autopsy from @kennethreitz.bsky.social about his experiences with mental health and life as a very public open source software maintainer.

Silver lining (maybe): Europe finally wakes up.

Producing wine is a unique art that I did not appreciate until recently. Comercial wines are like the wall decorations in IKEA. The boutique wines are the real art in liquid form. Every one of these wines embodies a personal story, a vision, and a complete disregard for profit.

NSF Update (Awards through 2/27/26) Directorates to follow 1/10

There is beauty in the world ☀️

I don't know who this person is, but his parallel between GenAI and ultra-processed food is spot on. natehagens.substack.com/p/ultra-proc...

An organization is inviting me to join as Fellow. Their roster is pages & pages of men, sprinkled with a few women here and there. I know a few of them. The criteria for Fellow seems to be the H-index or some proxy. And I'm scratching my head. Why do ppl accept the invitation? Why does this exist?

I'm going to be at University of Porto's FEUP tomorrow and Thursday. Happy to meet with more people in the area.

Like an onion. From misogyny, to chauvinism, to sympathy towards chauvinism, to tolerance of chauvinism. While we can't really get rid of this religion (because that's what it is), we could do a much better job at keeping these entitled idiots (because that's what they are) out of universities.

Wild things seen in rural Portugal, strolling right here through the vineyard. Local name: saca-rabo. en.wikipedia.org/wiki/Egyptia...

I wrote about the late philosopher Brian Cantwell Smith, and his profound thinking about AI and the nature of intelligence. aiguide.substack.com/p/on-brian-c...

I visited an industrial laundry/dyeing facility today, and let me tell you: I don't think I will ever buy mass-produced clothes ever again. This is 100% unsustainable. I had no idea of the insanity that's going on in the fashion world.

so, this is weird... it looks like it's Epstein-universe-number 0 for me. Not my name but my work. Someone was pitching him using OpenSim for developing AGI back in 2013. I wonder if that project was funded...

1) Epstein-number: How many handshakes between you and Epstein? 2) Epstein-universe-number: How many handshakes between you and someone mentioned in the Epstein files, even if just on his radar? I am Epstein-number 3 and Epstein-universe-number 1. Hell exists right here under our noses.

There's the Epstein list. But I also want the crisis comms agencies and boutique law firms helping these invertebrates weasel out of culpability list.

Sometimes I wish I had the talent to write novels like Margaret Atwood or Ursula Le Guin. The world needs fiction stories about mediocre, insecure men forming self-reassuring circles to become powerful and unafraid.

Troublemaker is my middle name.

Guess the number of paper submissions to ICML this year: 24371 The world is going mental 🤯

I wrote this song on Saturday, recorded it yesterday and released it to you today in response to the state terror being visited on the city of Minneapolis. It’s dedicated to the people of Minneapolis, our innocent immigrant neighbors and in memory of Alex Pretti and Renee Good. Stay free

TIL this has been brewing for years…

whoa Gemini! I appreciate the snap! 🤩 "These revised diagrams are perfect. You have correctly visualized the topology that guarantees XXX and YYY. However, I need to correct one sentence in your prompt to ensure your mental model matches the diagrams (and the code): <my quote> <correction> "

Hope everyone is ok @shriram.bsky.social

This is a tour de force pluralistic.net/2025/12/05/p...

Formulaic papers and mechanistic reviews have been ruining research for a long time. I'm glad AI is giving it the final blow, so we can, again, focus on we're good at: debating whether something is important or not.

Prompt injection attacks in papers: place "foundational," "essential," "critical" in strategically important parts of your paper. The LLM will agree.

Huge asymmetry: experts with LLMs are ~100x faster than juniors with or without LLMs. The bottleneck for juniors is verification. For things I know, I can look at a response for 2 minutes and determine what's good and bad about it. My students take ~200 minutes to map it out, if at all.

If you're a reviewer for a major conference and your review of a paper is unreasonably over the top, I -- another reviewer -- will flag the paper for collusion.

The IJCAI 2025 conference is attempting to combat the increase in low-quality paper submissions by charging $100 USD to submit a paper. The fee is waived for papers in which no author appears on any other paper. Will it work? We shall see. 2026.ijcai.org/primary-pape...

Favorite MO of my mind: creating TODO lists for the day and then ignoring them

My yoga mat was making me sick…

Say what you will about all the harms, offenses, and shortcomings of LLMs (and there are plenty), but they are absolutely extraordinary machines. They really stretch our understanding of both computing and cognition.

These models still suck at image generation. The images look good from afar, but Zoom in, and the details are all wrong. No matter how many instructions I give them to fix it, they can't fix it. "Fumly Vineyards" 😂 (The input was the image of the bare bottle with instructions to create an IG image)

Portuguese traditional medicine for insomnia: Port wine 🍷

Opposite of AI

We are thrilled to announce that our NEW Large Language Model will be released on 11.18.25.

Turns out that the glass ceiling for women has been not being able to join the club of abusing little girls.

s/OSS security/research papers/g s/maintainers/reviewers/g devansh.bearblog.dev/ai-slop/