iiinova — where your digital self keeps evolving

chenhaotan.bsky.social @chenhaotan.bsky.social · 23/08/2026

It is very hard to keep up with multiple social media on multiple platforms. Allow me to share some old news. When ICML wrapped up, we used AI agents to review and replicate all 168 oral papers, including 105 full replications. The results are in this blog: sai.science/blog/how-muc...

sai.science

How much science is verifiable? Results from replicating ICML 2026 oral papers

When ICML wrapped up, we used AI agents to review and replicate every oral paper to see what we can learn from actually verifying the science.

0252

chenhaotan.bsky.social @chenhaotan.bsky.social · 23/08/2026

I have been thinking along the same line. This is what we need.

061

Reposted by @chenhaotan.bsky.social

Data Science Institute @dsi-uchicago.bsky.social · 17/03/2026

It’s Election Day 🗳️ UChicago's @chenhaotan.bsky.social and team built CivicChats to help voters engage more thoughtfully with the issues on their ballot. Try the tech: civicchats.org Read more: datascience.uchicago.edu/insights/cou...

civicchats.org

CivicChats

A platform for exploring, debating, and thinking through upcoming ballot measures on your local ballot or across the U.S.

011

Reposted by @chenhaotan.bsky.social

Paperpile @paperpile.com · 11/03/2026

Open AI-assisted review tools could help researchers get better feedback, improve paper quality, and make the review process more scalable. But final publication decisions should remain under human oversight, via @chenhaotan.bsky.social openaireview.github.io/blog.html

openaireview.github.io

AI-assisted Reviewing is Necessary and Should be Open

Peer review is facing a death spiral. AI production tools are speeding it up. AI-assisted reviewing is necessary and should be open.

021

chenhaotan.bsky.social @chenhaotan.bsky.social · 10/03/2026

The blog actually discussed different types of reviews. If AI reviewing helps authors produce better science, I do not see why one needs to be so hostile against AI. It actually helps authors slow down to produce better-quality articles.

100

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

Try it, read the blog, or contribute: 🌐 openaireview.github.io 📝 openaireview.github.io/blog.html 💻 github.com/ChicagoHAI/O...

openaireview.github.io

OpenAIReview — AI-Powered Academic Paper Reviewer

010

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

We also don't have good evaluations for AI-generated reviews yet. We're working on it and welcome collaborators. Feedback welcome, especially from conference organizers and journal editors who want to think seriously about the future of peer review.

110

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

There are two types of reviewing. Reviewing for quality (improving the work) — what Refine and OpenAIReview do — is very different from gatekeeping (accept/reject), which is what Stanford Agentic Reviewer targets. We think automating gatekeeping requires much more care.

110

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

Our progressive approach finds issues at 87% of locations flagged by Refine, for the price of a coffee per paper. @joehsu.bsky.social added a Claude skill making it essentially free for Claude subscribers.

110

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

3/7 The only intervention that can stabilize the system is improving review precision, the ability to distinguish good papers from weak ones. AI production tools lower submission costs; only AI review tools can raise the signal. That asymmetry is why we built OpenAIReview.

100

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

2/7 The review death spiral: more submissions → overloaded reviewers → noisier reviews → more random acceptance → even more submissions. Bergstrom & Gross already warned about this. AI production tools make it worse by lowering submission costs and pushing the system toward collapse faster.

110

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/03/2026

Peer review is facing a death spiral, and AI production tools are speeding it up. AI-assisted reviewing is necessary and should be open. We built OpenAIReview: open AI reviewing for everyone, for the cost of a coffee. openaireview.github.io/blog.html 🧵

openaireview.github.io

AI-assisted Reviewing is Necessary and Should be Open

Peer review is facing a death spiral. AI production tools are speeding it up. AI-assisted reviewing is necessary and should be open.

12610

Reposted by @chenhaotan.bsky.social

Mourad Heddaya @mheddaya.bsky.social · 25/02/2026

Local ballot measures are now on CivicChats! Local elections happen year-round, 10+ states have measures coming up in the next few months. Check your ballot and think through what you'll be voting on → civicchats.org

021

chenhaotan.bsky.social @chenhaotan.bsky.social · 20/02/2026

We have been developing automatic evaluation based on checklists. We are also planning to run a study at the same time. Learn more at the end of this blog: cichicago.substack.com/p/civicchats...

cichicago.substack.com

CivicChats - Building AI to support voting behavior

CivicChats is a platform for exploring, debating, and thinking through upcoming ballot measures.

000

chenhaotan.bsky.social @chenhaotan.bsky.social · 19/02/2026

Check out our effort in thinking about how AI can help with democratic processes!

111

chenhaotan.bsky.social @chenhaotan.bsky.social · 16/02/2026

Anyone can help reviewing an ACL submission today on parameter efficient fine-tuning? Sorry that it is very tight.

001

Reposted by @chenhaotan.bsky.social

Xiaoyan Bai @elenal3ai.bsky.social · 10/02/2026

📖 ≠ 🧪 The Story is Not the Science. Code is submitted but rarely executed during peer review—an issue likely to worsen with research agents. 🧑‍🔬 We introduce 𝐌𝐞𝐜𝐡𝐄𝐯𝐚𝐥𝐀𝐠𝐞𝐧𝐭, an execution-grounded evaluation of narrative + execution. 𝐕𝐞𝐫𝐢𝐟𝐲 𝐭𝐡𝐞 𝐬𝐜𝐢𝐞𝐧𝐜𝐞, 𝐧𝐨𝐭 𝐣𝐮𝐬𝐭 𝐭𝐡𝐞 𝐬𝐭𝐨𝐫𝐲. 1/n

294

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/02/2026

Mark Yatskar will be speaking this Friday! You can tune in either on Zoom: uchicago.zoom.us/j/9897879984... Youtube: www.youtube.com/@AIScientifi...

000

chenhaotan.bsky.social @chenhaotan.bsky.social · 02/02/2026

Hannes Stark will be speaking this Friday on BoltzGen! You can tune in either on Zoom: uchicago.zoom.us/j/9897879984... Youtube: www.youtube.com/@AIScientifi...

000

Reposted by @chenhaotan.bsky.social

Hanna Wallach @hannawallach.bsky.social · 29/01/2026

Microsoft Research NYC is hiring a researcher in the space of AI and society!

26240

chenhaotan.bsky.social @chenhaotan.bsky.social · 26/01/2026

@profbuehlermit.bsky.social from MIT will be speaking this Friday! You can tune in either on Zoom: uchicago.zoom.us/j/9897879984... Youtube: www.youtube.com/@AIScientifi...

041

chenhaotan.bsky.social @chenhaotan.bsky.social · 20/01/2026

Peter Clark from @ai2.bsky.social will be speaking on Friday! You can tune in either on Zoom: uchicago.zoom.us/j/9897879984... Youtube: www.youtube.com/@AIScientifi...

030

Reposted by @chenhaotan.bsky.social

Paul Goldsmith-Pinkham @paulgp.com · 19/01/2026

Posted a very early stage draft with rock star collaborators. Key question: when we actually roll out AI tools, how do people use them? Do they just defer completely? Does it improve productivity and ability? We look in the medical setting of pulmonary embolisms paulgp.com/papers/Radio...

We study how radiologists use AI to diagnose pulmonary embolism (PE), tracking over 100,000
scans interpreted by nearly 400 radiologists during the staggered rollout of an FDA-approved
diagnostic platform. When AI flags PE, radiologists agree 84% of the time; when AI predicts no PE,
they agree 97%. Disagreement evolves substantially: radiologists initially reject AI-positive PEs in
30% of cases, dropping to 12% by year two. Despite a 16% increase in scan volume, diagnostic speed
remains stable while per-radiologist monthly volumes nearly double, with no change in patient
mortality—suggesting AI improves workflow without compromising outcomes. We document
significant heterogeneity in AI collaboration: some radiologists reject AI-flagged PEs half the time
while others accept nearly always; female radiologists are 6 percentage points less likely to override AI
than male radiologists. Moderate AI engagement is associated with the highest agreement, whereas
both low and high engagement show more disagreement. Follow-up imaging reveals that when
radiologists override AI to diagnose PE, 54% of subsequent scans show both agreeing on no PE
within 30 days.

48718

Reposted by @chenhaotan.bsky.social

Johan Ugander @jugander.bsky.social · 08/01/2026

I've often joked that as faculty I program in a high-level language called "graduate student". Having tried out Claude Code this morning, I (i) feel extremely at home, (ii) am realizing that research-by-graduate-student is perhaps the original vibe-coding. 1/2

78711

Reposted by @chenhaotan.bsky.social

David Jurgens @davidjurgens.bsky.social · 18/01/2026

I've seen this message and similar echos for other writing, and I want strongly push back on this narrative. It's not that you shouldn't use ChatGPT but that you shouldn't *use ChatGPT to write it for you*. ChatGPT—and AI in general—is not a monolith. How you use it matters.

181

chenhaotan.bsky.social @chenhaotan.bsky.social · 16/01/2026

Very much enjoyed this talk by @yisongyue.bsky.social ! The measurement challenge deserves a lot more attention from the AI community!

020

Reposted by @chenhaotan.bsky.social

Sophie Greenwood @sjgreenwood.bsky.social · 14/01/2026

Excited to present a new preprint with @nkgarg.bsky.social: presenting usage statistics and observational findings from Paper Skygest in the first six months of deployment! 🎉📜 arxiv.org/abs/2601.04253

417150

chenhaotan.bsky.social @chenhaotan.bsky.social · 15/01/2026

Emergent misalignment made into @nature.com! The key insight is that models fine-tuned on writing insecure code present a wide range of insecure behavior in other contexts.

040

chenhaotan.bsky.social @chenhaotan.bsky.social · 13/01/2026

I think it would be useful to attract researchers in industry to the platform as well.

030

Reposted by @chenhaotan.bsky.social

Data Science Institute @dsi-uchicago.bsky.social · 12/01/2026

Join us on Friday, January 16th for this week’s AI & Scientific Discovery Online Seminar! Featuring Yisong Yue, Professor of Computing and Mathematical Sciences at @caltech.edu You can participate online or in-person at DSI. Learn more at ai-scientific-discovery.github.io

021

chenhaotan.bsky.social @chenhaotan.bsky.social · 12/01/2026

Professor @yisongyue.bsky.social will be speaking this Friday at the AI & Scientific Discovery seminar! Zoom: uchicago.zoom.us/j/9897879984... Youtube: www.youtube.com/@AIScientifi...

020

chenhaotan.bsky.social @chenhaotan.bsky.social · 12/01/2026

I certainly did not realize that this is such a controversial take: Chenhao: "Finding the equilibrium of publishing will take at least a decade." 25% agree, 75% disagree

100

Reposted by @chenhaotan.bsky.social

George Conway ⚖️🇺🇸 @gtconway.bsky.social · 12/01/2026

Let us all stand with Chairman Powell.

968331798659

Reposted by @chenhaotan.bsky.social

David Bau @davidbau.bsky.social · 12/01/2026

I can't read Chinese, but my family has old genealogy documents I've always wanted to understand. Claude and Gemini helped me build an interactive reader to explore the calligraphy character by character. I can finally read my great-grandfather's epitaph. Try it: davidbau.com/archives/202...

Screenshot of Chinese calligraphy reader web application

0254

chenhaotan.bsky.social @chenhaotan.bsky.social · 11/01/2026

in general yes, but yonatan requested this not be recorded.

100

Reposted by @chenhaotan.bsky.social

Jeff Dean @jeffdean.bsky.social · 09/01/2026

The recent days have been horrific. We can't become numb to repeated instances of illegal and unconstitutional action by government agencies. It's even worse when public officials are blatantly lying in ways that contradict dozens of pieces of video evidence.

519825

Reposted by @chenhaotan.bsky.social

ACM FAccT @facct.bsky.social · 10/01/2026

We have an important #FAccT2026 update! Our new dates are June 25-June 28 and our new location is Le Centre Sheraton Montréal. We’re so sorry for any inconvenience caused by this change! Why the change? Thread --> 1/n

197

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/01/2026

The first talk is happening in two hours! Yonatan Belinkov is going to talk about "Interpretability for Scientific Discovery: Some Opportunities and Challenges"! You can tune in either on Zoom: uchicago.zoom.us/j/9897879984... Youtube: youtube.com/@AIScientifi... Hope to see you soon!

uchicago.zoom.us

Join our Cloud HD Video Meeting

Zoom is the leader in modern enterprise cloud communications.

130

chenhaotan.bsky.social @chenhaotan.bsky.social · 09/01/2026

Happy new year! The AI & Scientific Discovery Seminar is returning this quarter. Last quarter was incredible, from protein design to automated bio labs. Huge thanks to all our amazing speakers and attendees 🙌 We’re kicking off Winter Quarter with an 🔥 lineup, starting in two hours! 🧵

131

Reposted by @chenhaotan.bsky.social

Alexandria Ocasio-Cortez @aoc.bsky.social · 03/01/2026

It’s not about drugs. If it was, Trump wouldn’t have pardoned one of the largest narco traffickers in the world last month. It’s about oil and regime change. And they need a trial now to pretend that it isn’t. Especially to distract from his sinking under Epstein and skyrocketing healthcare costs.

1170314388339

chenhaotan.bsky.social @chenhaotan.bsky.social · 03/01/2026

It is tough. I wish there is more content here, but there is a lot of useful AI-related content on the other platform.

150

Reposted by @chenhaotan.bsky.social

Joseph M. Adelman @jmadelman.bsky.social · 14/12/2025

Brown has canceled all remaining classes and exams for the Fall semester.

Email from Provost Francis J. Doyle
Subject: Update on Remaining Fall 2025 Classes and Exams

Dear Members of the Brown Community,

I'm writing to follow up on President Paxson's update from a short time ago to provide additional information on academic operations for the campus. Given the deeply tragic events that took place on our campus yesterday, all remaining undergraduate, graduate and medical classes, exams, and papers or projects for the Fall 2025 semester will not take place as scheduled. This choice was made out of our profound concern for all students, faculty and staff on our campus. In the immediate aftermath of these devastating events, we recognize that learning and assessment are significantly hindered in the short term and that many students and others will wish to depart campus. Students are free to leave if they are able. Students who remain will have access to on-campus services and support.

135298