David Mimno
dmimno.bsky.social
David Mimno
@dmimno.bsky.social
He teaches information science at Cornell. http://mimno.infosci.cornell.edu
Pinned
Optimist: AI has achieved human level performance!

Realist: “AI” is a collection of brittle hacks that, under very specific circumstances, mimic the surface level of human intelligence

Pessimist: AI HAS achieved human level performance
Could a data center moratorium help US tech? DeepSeek took off because the lazy "just buy bigger GPUs" strategy wasn't available to them. Giving companies some relief from FOMO plus incentive to actually innovate could avoid an involution trap (exponential cost, log returns)
We updated the plots we use to measure the open model ecosystem at
interconnects, to guide The ATOM Project, and to understand what's happening.

We have ~8 plots to summarize what's happening.
First, the high level picture showing China's growing adoption lead.
January 7, 2026 at 4:41 PM
Reposted by David Mimno
a huge victory for universities, worth billions of dollars: Trump can’t cap indirect-cost rates on NIH grants, appeals court rules
www.chronicle.com/blogs/the-tr... via @andyandy.bsky.social
Trump Can’t Cap Overhead Rate on NIH Grants to Research Universities, Appeals Court Rules
The ruling is a victory for higher-education associations who challenged the proposed 15-percent cap, calling it illegal and claiming it would devastate the research enterprise.
www.chronicle.com
January 6, 2026 at 6:29 PM
There are many, many problems that are Not Hard and that we have excellent, cheap solutions for (calculations, first order logic, running python programs, chess). Language models don't need to do everything!
"GenAI is the futu--
January 6, 2026 at 2:08 PM
I used Claude to improve my arXiv API scraping code and topic display, here's an updated view of trends over time in the computer vision section: mimno.infosci.cornell.edu/arxiv/cv

(top rising topic: "Language")
CV Topics in arXiv - Interactive Visualization
Interactive visualization of 100 computer vision topics from arXiv cs.CV abstracts showing trends over time
mimno.infosci.cornell.edu
January 5, 2026 at 9:03 PM
Reposted by David Mimno
📢 Apply by Feb 12 to join our CHI 2026 workshop, Speech AI for All, where we'll discuss inclusive speech tech for people with speech diversities. Researchers, practitioners, policymakers, & community members welcome! speechai4all.org
January 5, 2026 at 5:50 PM
From the 2022 drop era: “who needs wheat? Bread is so much tastier”
I guess StackOverflow is done.
January 4, 2026 at 5:57 PM
Until I see evidence otherwise, I’m going to assume the “run the country” thing was just ad libbed
January 3, 2026 at 6:56 PM
Vardaman

My mother is a fish.
January 1, 2026 at 4:57 PM
This is one of the big design choices: can an AI-forward programming class be the first exposure, or is this always setting students up to fail without the basic variables-and-for-loops programmer-thinking class? That one is looking like mostly on-paper, going forward, btw.
hot take: an introductory class intended for students for whom it will be their first class in a sequence should remain essentiailly unchanged. The fundamentals are still the fundamentals. The difference should be analogous to how presentation of IDE features to intro classes has changed over time.
December 31, 2025 at 2:36 PM
What does a class on software development post LLM contain?

Design?
Syntax of common langs?
Programmer thinking?
Environments/libraries/bash?
Debugging?

References and texts especially welcome
December 30, 2025 at 8:10 PM
Pushed a bunch of long-overdue updates to Mallet using Claude Code. Updated to Java 17, moved tests to a separate directory, updated to JUnit 4, switched to all maven. Testers welcome!
December 29, 2025 at 5:04 PM
Gift is not a verb
December 24, 2025 at 4:15 PM
Reposted by David Mimno
PhD and Postdoc positions drop! The Research Council of Norway funded MishMash Centre for AI & Creativity has announced 18 doctoral and 6 postdoc positions in arts and/or science. I’m a member of the Scientific Advisory Board and so pleased to see this happening.

mishmash.no/news/positio...
18 PhD and 6 Postdoc positions at MishMash/Norway
The first round of MishMash doctoral and postdoctoral fellowships are now announced.
mishmash.no
December 22, 2025 at 5:31 PM
Reposted by David Mimno
We are looking for advanced PhD student interns! Spend 12 weeks scratching a research itch in our Cambridge Massachusetts lab. Closes January 17.

apply.careers.microsoft.com/careers/job/...
Research Intern - Sociotechnical Systems | Microsoft Careers
Research Interns put inquiry and theory into practice. Alongside fellow doctoral candidates and some of the world's best researchers, Research Interns learn, collaborate, and network for life. Researc...
apply.careers.microsoft.com
December 22, 2025 at 9:32 PM
In pre-20c literature a man will often ask a woman "to be her servant" ("he hit quyk askez / To be her seruaunt sothly"). What does this actually imply?
December 22, 2025 at 8:23 PM
Bots become more human-like, humans become more bot-like
It seems once one anti AI user gets ahold of a post, all of a sudden a ton of accounts attach themselves to the post.

It could very well be a bunch of human anti-AI haters. Totally plausible. To me, it seems like mostly bot behavior.

My post could be improved w/"there seems to be a bot network"
December 22, 2025 at 1:07 AM
Reposted by David Mimno
This program brought together such a wonderful group of ~30 PhD students for two weeks at Berkeley last year with backgrounds in sociology, information, law, computer science and more — same idea but in Oxford next summer; apply by Feb 2 if you’re looking to make connections across fields
December 19, 2025 at 11:46 PM
Encoder-decoder models are great for annotation tasks, where you want full information about an input as well as the ability to generate arbitrary structured output. Encoders (BERT) and Decoders (most everything else) only give you one of those.
Google released the open-weights of T5Gemma 2, an encoder-decoder models

Built on top of Gemma 3, T5Gemma 2 is a multimodal, long-context, and heavily multilingual (140 languages) encoder-decoder model, where most models today are decoder-only.
December 19, 2025 at 3:57 PM
Reposted by David Mimno
🏹 Job alert: Two fully funded PhD positions in Natural Language Processing at University of Leipzig

📍 Leipzig 🇩🇪
📅 Apply by Jan 15th
🔗 https://ellis.eu/research/jobs/2025-12-16-two-fully-funded-phd-positions-in-natural-language-processing
Two fully funded PhD positions in Natural Language Processing at University of Leipzig | European Laboratory for Learning and Intelligent Systems
ellis.eu
December 18, 2025 at 7:05 AM
Reposted by David Mimno
📣 Competition Launch Alert! Deep Past Challenge: Translate Akkadian to English hosted by Deep Past AI

🎯 Build an AI model that translates 4,000-year-old Old Assyrian business records into English
💰 $50,000 Prize Pool
⏰ Entry Deadline: March 23, 2026

www.kaggle.com/competitions...
Deep Past Challenge - Translate Akkadian to English
Bringing Bronze Age Voices Back to Life – Machine Translation of Old Assyrian Cuneiform
www.kaggle.com
December 17, 2025 at 1:57 PM
Use AI unto others as you would have them use AI unto you
December 16, 2025 at 5:54 PM
Reposted by David Mimno
COLM 2026 is just around the corner! Mark your calendars for:

💡 Abstract deadline: Thursday, March 26, 2026
📄 Full paper submission deadline: Tuesday, March 31, 2026

Call for papers (website coming soon):
docs.google.com/document/d/1...
December 16, 2025 at 3:31 PM
Goodnight sweet prince
December 15, 2025 at 2:17 PM
Rebecca Hicke presents our work on Christian fiction #CHR2025
December 12, 2025 at 1:07 PM
What does it mean for a song to be narrative? @dbamman.bsky.social Sabrina Baur, Mackenzie Hanh #CHR2025
December 12, 2025 at 10:51 AM