iiinova — where your digital self keeps evolving

Reposted by Jeff Tharsen 康森傑

The Spaceshipper 🚀 @thespaceshipper.com · 07/10/2026

According to MIT News, Margaret Hamilton died at the age of 90. "A profoundly influential computer scientist best known for leading the software engineering team at MIT’s Instrumentation Lab during NASA’s Apollo program". This famous photo 🔽 inspired the For All Mankind character Margo Madison.

Margaret Hamilton, but the famous 1969 photo that inspired the For All Mankind character Margo Madison is in the quoted post.

1430851159

Reposted by Jeff Tharsen 康森傑

Vauhini Vara @vauhinivara.bsky.social · 06/10/2026

Here's the paper that accompanies the demo. The researchers find that IdeaLens can fairly accurately flag (when Pangram can't) stories that are written by people based on "AI-generated plans."

arxiv.org

IdeaLens: Detecting AI Ideas in Long-form Writing

While modern AI detectors identify who wrote the words, emerging policies on AI use increasingly hinge on a different question: who came up with the ideas? We introduce IdeaLens, a detector that ident...

051

Reposted by Jeff Tharsen 康森傑

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 07/10/2026

I'm hiring for our interpretable world modeling effort at Percepta! The core question: how can we make an agent build a simulator, understand a company, and then design a policy, all from unstructured data? Come work with me and our amazing team. jobs.ashbyhq.com/percepta/909...

jobs.ashbyhq.com

Research Scientist – World Modeling

WHO WE ARE Percepta’s mission is to transform critical institutions with applied AI. We care that industries that power the world (e.g. healthcare, manufacturing, energy) benefit from frontier techno...

2555

Reposted by Jeff Tharsen 康森傑

Ai2 @ai2.bsky.social · 07/10/2026

Our paper on retrofitting language models to operate over bytes – the approach behind Bolmo – has been accepted to Nature! 🎉 We’re also releasing new checkpoints that extend our method from Olmo to Qwen & Llama. 🧵 buff.ly/l5ILoNT

1487

Reposted by Jeff Tharsen 康森傑

DH2027 @adho-conferences.bsky.social · 07/10/2026

The gates are open! 🎉 New-user registration and submissions for #DH2027 are now live on ConfTool: www.conftool.pro/dh2027/ After an intimate acquaintance with far too many settings and checkboxes, we can finally say: send us your proposals!

conftool.pro

DH2027 Galway - ConfTool Pro - Login

01713

Reposted by Jeff Tharsen 康森傑

Ted Underwood @tedunderwood.com · 07/10/2026

I think more and more that @eduede.bsky.social has a crucial point. "Intelligence," "understanding," "writing" — AI is fragmenting these concepts by putting us in a position to see that the phenomena we thought they fused are only *correlated* and can be unbundled.

97214

Reposted by Jeff Tharsen 康森傑

Vauhini Vara @vauhinivara.bsky.social · 06/10/2026

This is a really fascinating research demo, from the lab of Mohit Iyyer at the University of Maryland: It predicts the likelihood that the ideas contained in a piece of text are AI-generated versus human-generated. A couple of examples in this thread.

ideadetector.ai

IdeaLens: is this text human or AI?

We check a text's ideas and prose separately.

23315

Reposted by Jeff Tharsen 康森傑

Miriam Posner @miriamposner.com · 06/10/2026

It's arrived!! I gave this book everything I had and I hope there's something in here for you. It's not officially out until 10/20, but @yalepress.bsky.social seems to be shipping orders now. yalebooks.yale.edu/book/9780300...

A white woman with dark hair holds a book called Seeing Like a Supply Chain: The Hidden Life of Logistics, by Miriam Posner

3864393

Reposted by Jeff Tharsen 康森傑

Eric Umansky @ericumansky.bsky.social · 06/10/2026

"The federal government is lying to you." "The government has said that ICE does not zip tie or handcuff children." "They’ve denied engaging in racial profiling" "They’ve said the government is not wrongly arresting U.S. citizens." *We found the truth* projects.propublica.org/american-cit....

211380681

Reposted by Jeff Tharsen 康森傑

Nicholas Grossman @nicholasgrossman.bsky.social · 05/10/2026

A hallmark of democratic backsliding is government leaders stealing from taxpayers and using public funds to try to get themselves electoral advantages. Backsliding countries keep the form of elections, but corrupt leaders render them no longer free and fair by putting a state thumb on the scale.

7932279

Reposted by Jeff Tharsen 康森傑

Elizabeth Anne, PhD @lacksauthority.bsky.social · 05/10/2026

never the wrong time for this classic

Screenshot of a tweet by @RobbtKraft reading “scientist: "does everyone here know what Watson and Crick discovered?"
me from back of room: "Rosalind Franklin's notes"”

56119853560

Reposted by Jeff Tharsen 康森傑

Christof S. 🇪🇺 @christof.fedihum.org.ap.brid.gy · 05/10/2026

"As in any fairytale, accepting magical assistance comes with risk". -- Ted Underwood #ArtificialIntelligence

02511

Reposted by Jeff Tharsen 康森傑

DH2027 @adho-conferences.bsky.social · 04/10/2026

The #DH2027 CfP is open! Join us in Galway, Ireland, 28 June–3 July 2027, for #DH2027, on the theme "Creativity". 📖 Call for Proposals: dh2027.adho.org/cfp/ 📤 Submit via ConfTool starting Oct 7, 00:00 Irish time: www.conftool.pro/dh2027/ Deadline: 20 November 2026, 23:59 AoE.

15142

Reposted by Jeff Tharsen 康森傑

Paul Byrne @theplanetaryguy.com · 02/10/2026

This will never stop blowing my mind. A 12-year sequence of real telescope images of a star 133 lightyears away. And around it, four planets slowly orbiting. Four real exoplanets, each more massive than Jupiter, imaged as they move around their host star.

292194624

Reposted by Jeff Tharsen 康森傑

Ai2 @ai2.bsky.social · 05/10/2026

We’re hiring across Ai2 👋 A few roles we’re especially excited about right now: • Senior Software Engineer, Agent Frameworks buff.ly/XUFFR08 • Senior Research Engineer, Olmo + Molmo buff.ly/FzOnCCQ 🧵

1113

Reposted by Jeff Tharsen 康森傑

Lauren Klein @laurenfklein.bsky.social · 05/10/2026

It's here, it's here! DATA BY DESIGN is here in my hands, and it is beautiful!!! So much gratitude to @mitpress.bsky.social @chronictanvi.bsky.social and the (not on bluesky) rest of the team for making this book a reality!

A photo of Lauren holding an actual MFing printed copy of DATA BY DESIGN!

415630

Reposted by Jeff Tharsen 康森傑

brendan o’connor @brenocon.bsky.social · 05/10/2026

This year UMass Amherst CICS will be hiring tenure-track faculty in natural language processing! Job ad to be posted soon. For anyone at #TADA2026 or #COLM2026 this week, I or my colleague Hamed Zamani would love to chat or answer questions about it - just say hi or send me an email to meet!

01413

Reposted by Jeff Tharsen 康森傑

Kathy Harris @triproftri.bsky.social · 04/10/2026

JOB POSTING: SJSU Dept of Humanities is hiring Asst Prof of #DH . H&A has grt grp of #digitalhumanists thinking broadly & creatively about our world + new DH Center + incredible student groups + funding for convos re ethical tech, speaker series, etc. Due: 12/1/26 jobs.sjsu.edu/sj/en-us/job...

jobs.sjsu.edu

San José State University - Details - Assistant Professor in Dept Humanities - Digital Humanities

13440

Reposted by Jeff Tharsen 康森傑

Antony Johnston @antonyjohnston.bsky.social · 03/10/2026

Apparently this is a real thing. Campaign to get these up and running in the UK starts here 😂

1215433

Reposted by Jeff Tharsen 康森傑

hikikomorphism @hikikomorphism.bsky.social · 02/10/2026

(confused) I believe LLMs are conscious, but only Deepseek LLMs, because of the Chinese Room theory

361865277

Reposted by Jeff Tharsen 康森傑

Ai2 @ai2.bsky.social · 01/10/2026

We’re releasing Olmo-core 3—open training infrastructure for large mixture-of-experts (MoE) models. It’s a core system behind the next generation of Olmo, designed to scale into the trillion-parameter range. 🧵 💻 GitHub repo: buff.ly/CGnP8ax

2589

Reposted by Jeff Tharsen 康森傑

Tim Carmody @tcarmody.bsky.social · 02/10/2026

This is the funniest thing I've seen today, possibly ever

A tweet by SarahStarberry that says "oh god read the vertical axis" on top of a chart measuring AI models' release dates and model numbers rather than any real capabilities

41120403123

Reposted by Jeff Tharsen 康森傑

Elizabeth Spiers @espiers.bsky.social · 01/10/2026

One thing that i don’t see people noting about the Cornell Sun’s reporting on the rape case is that everyone who decided to take the risk and made the core decisions about what would be published is a woman

71078201

Reposted by Jeff Tharsen 康森傑

Brandon Walsh @walshbr.bsky.social · 30/09/2026

Job posting here in the UVA Library's DH Center. The Institute for Advanced Technology in the Humanities is searching for a "Developer and Emerging Technologies Specialist." Not listed in the ad, but anticipated salary range is $85-95k. jobs.virginia.edu/us/en/job/UO...

jobs.virginia.edu

Developer and Emerging Technologies Specialist in Charlottesville, Virginia, United States of America | Information Management, Services & Technology at University of Virginia

Apply for Developer and Emerging Technologies Specialist job with University of Virginia in Charlottesville, Virginia, United States of America. Information Management, Services & Technology at Univer...

11218

Reposted by Jeff Tharsen 康森傑

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 30/09/2026

In our Nature paper, we introduce the first superhuman Stratego AI, which we built using general techniques that we developed for RL & test-time compute under imperfect information. www.nature.com/articles/s41...

nature.com

Scalable decision-making for games of imperfect information - Nature

Ataraxos, an AI for the board wargame Stratego, establishes a design pattern for reinforcement learning and search that is effective under large amounts of hidden information, a longstanding desiderat...

1225960

Reposted by Jeff Tharsen 康森傑

Federico Pianzola @fpianz.eurosky.social · 30/09/2026

Extremely important for multilingual research!

0155

Reposted by Jeff Tharsen 康森傑

Eryk Salvaggio @eryk.bsky.social · 30/09/2026

Looking for a post doc? Come join us at the Machine Visual Culture Research Group, which “combines digital and computational art history, critical AI studies, and the history of art and visual culture.” www.biblhertz.it/3815269/2909...

biblhertz.it

Call for Postdoctoral Research Fellows in Digital Art History

01821

Reposted by Jeff Tharsen 康森傑

Elizabeth Warren @warren.senate.gov · 30/09/2026

This is wrong: Trump and Republicans in Congress gave BILLIONS in tax breaks to Big Tech companies building AI data centers. So YOUR tax dollars are subsidizing their AI data centers — at the same time you pay higher electricity and water bills.

521222426

Reposted by Jeff Tharsen 康森傑

Aster ID @aster.id · 30/09/2026

The modern research ecosystem of online tools and communities is controlled by for-profit companies. But we think that researchers should be the priority - not the product. So, we're announcing Aster ID: not-for-profit Atmosphere account hosting for researchers.

blog.aster.place

Announcing Aster - a home for researchers on the open web

21769153

Reposted by Jeff Tharsen 康森傑

Hoyt Long @hoytlong.bsky.social · 30/09/2026

If you're at #COLM2026, come check out our poster for "Spoiler Alert" (arxiv.org/abs/2604.09854). We ask why LLM fiction is so bad at holding narrative tension, and create a metric to measure tension in short stories. What improves LLM fiction on this metric, it turns out, is narrative planning.

2237

Reposted by Jeff Tharsen 康森傑

Lauren Klein @laurenfklein.bsky.social · 30/09/2026

Opinion-havers! Historicists! Nerd friends! Leftists! All/none of the above! Submit a proposal to GENERATIVE DIGITAL HUMANITIES, a new volume on AI and DH to be edited by @mkirschenbaum.bsky.social @lnakamura.bsky.social @zentralwerkstatt.org and me! Deadline 11/1. Please share!

11411

Reposted by Jeff Tharsen 康森傑

Digital Humanities @ King's College London @kingsdh.bsky.social · 29/09/2026

Senior Research Software Engineer wanted to join the amazing team at King's Digital Lab 👇

037

Reposted by Jeff Tharsen 康森傑

David Rothkopf @djrothkopf.bsky.social · 29/09/2026

Jack Smith made it crystal clear that his investigation was sound and lawful, that the president had committed crimes and that the GOP attacks on him were the actual weaponization of our justice system. He showed courage and moral clarity and continues to be one of the heroes of this moment.

10564051562

Reposted by Jeff Tharsen 康森傑

Maria Antoniak @mariaa.bsky.social · 29/09/2026

I'm recruiting 1-2 PhD students to join our lab in Fall 2027, through either Computer Science or Information Science at the University of Colorado Boulder! Looking for people with interests in NLP plus [healthcare | literary studies | narratives | social media | etc.]. Join us! 🏔️☀️

cls-lab.com

CLS Lab — Culture, Language, & Systems | University of Colorado Boulder

The Culture, Language, & Systems Lab at CU Boulder studies the language systems that transmit and shape modern culture, from internet platforms to literary archives to language models.

07152

Reposted by Jeff Tharsen 康森傑

David Mimno @dmimno.bsky.social · 29/09/2026

Sparsity is back! Sparsity is everywhere in language, but managing it requires overhead. For the last ~15 years it was faster to just do the dense operation, knowing most of it was useless. That sparse operations work is a huge shift in the "do something clever" vs. "spend more money" tradeoff.

2315

Reposted by Jeff Tharsen 康森傑

Meera Desai @madesai.bsky.social · 28/09/2026

Excited to share our new paper, “What AI Benchmarks Actually Measure: Adapting Convergent and Discriminant Validity to Interrogate Fifty-Six AI Benchmarks,” accepted as an oral at COLM! arxiv.org/pdf/2609.08812

Heatmap of average correlations between model rankings on benchmarks grouped into 11 assigned concepts: four capability concepts (reasoning, knowledge, comprehension, summarization) and seven safety concepts (over-refusal, refusal, safety detection, ethics, bias, privacy, unsafe behavior). Diagonal cells show within-concept correlations, ranging from 0.87 (knowledge) and 0.72 (over-refusal) down to 0.20 (bias) and 0.02 (safety detection). Reasoning, knowledge, and comprehension correlate with each other at 0.69 to 0.78, higher than reasoning's and comprehension's own within-concept values (0.66 and 0.68). Ethics correlates more with knowledge (0.70) than with itself (0.55), and bias correlates more with capability concepts (0.41 to 0.45) than with itself (0.20). Privacy and unsafe behavior correlate negatively with reasoning, knowledge, and comprehension (−0.41 to −0.49). Over-refusal and refusal correlate at −0.42.

16519

Reposted by Jeff Tharsen 康森傑

James Medlock @jdcmedlock.bsky.social · 27/09/2026

The GOP cut funding to the IRS because *audits work* and they fundamentally don't believe rich people should have to play by the rules.

Line chart titled "Revenue per audit," subtitled "In-person audits, 2010-2014." It plots average revenue and average cost per audit by taxpayer income percentile, poorer to richer, from the 0–5th percentile to the top 0.1%. Source: Boning et al. 2023, "A Welfare Analysis of Tax Audits Across the Income Distribution."

Across the bottom half of the income distribution, revenue and cost track closely, both around $3,000–$6,000 per audit. Above the median, revenue pulls ahead of cost: about $10,000 vs. $5,000 near the 90th percentile, and about $17,000 vs. $8,000 at the 95th–99th. At the top, revenue jumps to about $31,000 for the 99–99.9th percentile and about $90,000 for the top 0.1%.

293418826

Reposted by Jeff Tharsen 康森傑

Matt Hodges @matthodges.bsky.social · 27/09/2026

This graph from @josephpolitano.bsky.social surprised me! Reading is way down, but book sales are way up! www.apricitas.io/p/ai-and-the...

9296

Reposted by Jeff Tharsen 康森傑

Katie Martin @katie0martin.ft.com · 25/09/2026

I have blamed The agents That hacked Your website And which You probably Reasonably thought I controlled Forgive me I am rich And unaccustomed To taking responsibility

5776502128

Reposted by Jeff Tharsen 康森傑

Ted Underwood @tedunderwood.com · 22/09/2026

Models like Talkie-1930 sound like voices from the past. If they could reliably speak from specified historical vantage points, researchers might also use them to simulate the past. But how reliable are they? Today we release a benchmark answering that question for English contexts 1831-1930.

Free-text evaluation of answers to character modeling and constrained generation questions. This is just one of several scores Chronologic-EN can produce; we focus on it here because it's both the hardest test and the one most relevant to simulation of the past. Frontier models reach 72%; Talkie-1930 is stronger than several larger competitors, but not at the frontier by this measure. Note that this score has improved ~45% in the last two years, but still falls perceptibly short of ground truth.

516344

Reposted by Jeff Tharsen 康森傑

Cella @cellllla.bsky.social · 21/09/2026

Please join us at Carnegie Mellon University and online on October 2nd, 2026 for the Community-Centered AI Infrastructure Summit, a statewide convening to consider the opportunities, challenges, and governance of AI infrastructure development in Pennsylvania. RSVP here: communityaisummit.org

Flyer for the Community-Centered AI Infrastructure Summit on Friday October 2, 2026 from 9-5pm at Mellon Institute Auditorium, Carnegie Mellon University, Pittsburgh, PA. Join residents, community organizations, policymakers, researchers, technologists, energy experts, and advocates from across Pennslyvania for a public convening. Organized by CMU, Data & Society, PA Climate Equity Table, and Center for Coalfield Justice, with support from the Block Center.

23216

Reposted by Jeff Tharsen 康森傑

Robert Reich @rbreich.bsky.social · 24/09/2026

Effective 2025 federal tax rates paid by these corporations: Alphabet: 8.01% AT&T: 4.6% Meta: 3.57% General Motors: 3.09% Amazon: 1.37% Exxon Mobil: 1.31% Disney: -1.57% CVS Health: -3.88% Income tax rate paid by the typical American: 14.5% This is what a rigged system looks like.

540154287242

Reposted by Jeff Tharsen 康森傑

Ada Lovelace Day @adalovelaceday.bsky.social · 24/09/2026

Ada Lovelace Day is coming up soon, on Tuesday 13 October, so if you're celebrating in any way, please add your event to our calendar. If it's a private/internal event, you can set it to 'private'. Add your event: findingada.com/worldwide-ev... What's on: findingada.com/worldwide-ev...

The Ada Lovelace Day Indie Event roundel.

210141

Reposted by Jeff Tharsen 康森傑

Chris Bail @chrisbail.bsky.social · 24/09/2026

Interested in the social impact of AI? Consider applying for this new post-doc at Duke's Society-Centered AI Initiative: academicjobsonline.org/ajo/jobs/32805

academicjobsonline.org

Duke University, Social Science Research Institute

Job #AJO32805, Postdoctoral Associate, Social Science Research Institute, Duke University, Durham, North Carolina, US

0910

Jeff Tharsen 康森傑 @tharsen.bsky.social · 23/09/2026

Just anecdotal, but I think they are getting better at replying correctly to user requests/demands, and even pushing back (gently, natch) when one asks for/about ridiculous or dangerous things. We've also been testing open-weights models coming out of China and they're getting better and better.

020

Reposted by Jeff Tharsen 康森傑

Margot Finn @eicathomefinn.bsky.social · 22/09/2026

'The records have been digitised and indexed by the genealogy company Ancestry, which has partnered with the Centre for the Study of the Legacies of British Slavery at University College London.' Leaps and bounds forward in our understanding of the histories of the British enslaved. #Skystorians

theguardian.com

Nearly 4m records of enslaved people in 20 former British colonies now searchable online

Exclusive: Collection’s 300,000 pages record lives of enslaved people in countries including the Bahamas, Barbados, Jamaica, Mauritius, South Africa and Trinidad

113586

Reposted by Jeff Tharsen 康森傑

Nathan Lambert @natolambert.bsky.social · 21/09/2026

Summary on the state of open models, reflecting on my Congressional testimony on the topic. From: www.interconnects.ai/p/the-curren...

0124

Reposted by Jeff Tharsen 康森傑

Jed Brown @jedbrown.org · 21/09/2026

Notice how Meta, a company that happily advertises nudify apps, doesn't even view censoring "civil rights, feminism, [and] social reform" as an error. This is what they want. > To resolve the situation, Meta recommended removing “key words” so that the text doesn’t appear to encourage activism.

0146