Anand Sarwate (he/him)
banner
ergodicwalk.bsky.social
Anand Sarwate (he/him)
@ergodicwalk.bsky.social
Randomly transitioning between information theory, ML, privacy, statistics, and other destinations. Not using \mathbb in a logo.
Got an email/newsletter using Lena as an example image. It's 2026. C'mon, people.
February 6, 2026 at 5:14 PM
Someone should gift my senator, Chuck Schumer, a copy of Max Frisch’s Biedermann und die Brandstifter.

I am guessing he does not know German, but I can’t find a translation in Spinelesscowardese so maybe English will have to do.

Perhaps he will need some Cliff‘s Notes.
January 27, 2026 at 6:22 AM
In case anyone is on the academic job market and has an upcoming online video interview:
* do some research on the department: what classes can you teach, who can you collaborate with?
* be able to answer "why do you want to come here" with something more concrete than "because you are great..."
January 27, 2026 at 12:49 AM
ICLR AC: "The scores were too low, reject. Reviewer A's score was the lowest."
Also ICLR AC: "Reviewer A was marked as AI generated."
Also also ICLR AC: "Reviewer A is unlikely to raise their score."

It's ok to reject the paper, but come on...
January 26, 2026 at 3:43 PM
My partner @drnycnerd.bsky.social and co-authors from @cnti.bsky.social published a report on how people use LLM-based chatbots to get the news (even if it is wrong).

cnti.org/reports/chat...
Action, Ease & Personalization: AI Chatbot News Experiences
How habitual AI chatbot users in the US and India stay informed
cnti.org
January 22, 2026 at 1:13 PM
“When we use terms like ‘output,’ ‘feedback,’ ’interface,’ and ‘networking’ to express the transactions of everyday life, we reveal how thoroughly artificial things now shape our sense of human being.” -Langdon Winner, “The Whale and the Reactor” (Preface)
January 3, 2026 at 2:43 AM
Bezos calling his new startup Prometheus just makes me think of Jay Sherman's college film. I'll just call it "Promotheus" from now on.
The Critic - "L'Artiste est Morte"
YouTube video by RoflRadicalRaddish
www.youtube.com
November 17, 2025 at 2:23 PM
Reposted by Anand Sarwate (he/him)
JD Vance claims that diversity weakens unions, as people end up distrusting each other and not organizing.

Let me tell you two menswear stories related to this claim. 🧵
October 30, 2025 at 8:07 AM
TFW you open a new browser tab and your fingers automatically type "overleaf.c" + tab.
September 30, 2025 at 1:18 AM
Reposted by Anand Sarwate (he/him)
Good news about the cutting-edge AI tools from the $3-trillion-dollar company.
September 2, 2025 at 12:10 AM
Reposted by Anand Sarwate (he/him)
Worn, vintage yet precious to my heart. Souvenir: a mousepad from #ISIT 1998, the International Symposium on Information Theory, in honor of Claude E. Shannon 1948 paper, "A mathematical theory of communication", where I was a lucky voluntary student.
September 1, 2025 at 4:53 PM
This. Actually this.

www.argmin.net/p/the-banal-...
The Banal Evil of AI Safety
Chatbot companies are harmful and dishonest. How can we hold them accountable?
www.argmin.net
August 28, 2025 at 8:39 PM
Reposted by Anand Sarwate (he/him)
by applying a QR decomposition, any matrix can be written as a QR code multiplied by an upper triangular matrix
July 14, 2025 at 4:06 PM
Reposted by Anand Sarwate (he/him)
I've created a new AI assistant called Bartleby that politely refuses to do any task you ask it to do.
April 29, 2025 at 9:39 PM
Reposted by Anand Sarwate (he/him)
Went to go look at Grok but got distracted by this quite funny burn on Continental philosophy.
July 9, 2025 at 4:20 AM
Reposted by Anand Sarwate (he/him)
The New York Times was more skeptical of the academic qualifications of a black woman who had secured tenure and become the president of Harvard than they were of a white man who washed out of a grad program with a massive scandal and then became a white nationalist troll.
The saga of the NYT's anti-Mamdani story taken from notorious white supremacist Jordan Lasker (they call him an "academic") gets even more amazing: Lasker's only impactful paper, which claimed white people are smarter, is infamous: it misused data so badly it got his tenured co-author fired. 2/
July 5, 2025 at 5:04 PM
Reposted by Anand Sarwate (he/him)
The saga of the NYT's anti-Mamdani story taken from notorious white supremacist Jordan Lasker (they call him an "academic") gets even more amazing: Lasker's only impactful paper, which claimed white people are smarter, is infamous: it misused data so badly it got his tenured co-author fired. 2/
July 4, 2025 at 3:26 AM
OpenReview's servers appear to be borked so I can't upload my already written reviews and I've planned to be off the next few days. NeurIPS appears to be very dire about deadlines but also don't have the infrastructure up.
July 3, 2025 at 1:06 AM
Reposted by Anand Sarwate (he/him)
While I've expressed a number of disagreements w Mamdani, Sen. Gillibrand's lacerating and tendentious attacks on him just cement my longstanding belief that's she's a self-seeking opportunist of the worst order. I fully respect those who think Al Franken was right to resign. But Gillibrand ...
June 29, 2025 at 1:25 AM
Did you know that ICML is also the International Conference on Malignant Lymphoma?

The "AI is a cancer" jokes just write themselves!

www.icml.ch/icml/home.html
Home - ICML
www.icml.ch
June 18, 2025 at 7:48 PM
It is not yet fall but already my leaves runneth over.

(Or amok?)
June 4, 2025 at 3:30 PM
Reposted by Anand Sarwate (he/him)
They're terminating OPT.

Today the nominee to head USCIS stated directly: he will end Optional Practical Training, the single largest channel for high-skill immigrants to work in the US.

6 weeks ago I warned about this & summarized the research on impacts of this drastic action—> @piie.com
Skilled immigration on the chopping block? Effects of eliminating "Optional Practical Training" in the US
The White House has taken numerous steps to sharply reduce the number of immigrant workers, with or without legal status, in the US economy. These policies, so far, typically target immigrants with le...
www.piie.com
May 21, 2025 at 9:04 PM
Visiting Bristol Uni for the week and I'm glad I'm in the Fry building (maths) and not the Abattoir.

... although I guess one might get fried after visiting the abattoir?
May 19, 2025 at 3:36 PM
The trend of creating "acronyms" (I call them nacronyms) by picking random letters from the *middle* of words in the system's name is bad enough, but somehow C turned into a K as well? Don't try so hard, people.

Actually, pulling random letters and calling it "meaning" is a good metaphor for AI.
Learned from Scott Newstok of this study of large language models that evaluates their use for "1) close reading as a form of reasoning … 2) college-level knowledge in the literary domain & 3) figurative language understanding as ... reading comprehension” arxiv.org/abs/2505.09825
May 19, 2025 at 9:48 AM
Excited to learn from @spmontecarlo.bsky.social at LSIT!
May 15, 2025 at 1:31 PM