Feng Han

PhD Student @LSEGovernment| Information Politics| State-Society Relations| Political Communication

[Jan Zilinsky @janzilinsky]

I've now found several papers for political scientists choosing an LLM to annotate text data:

- Hilbig (2026) Open-Weight LLMs Are Often Competitive with Commercial APIs for Political Science Text Classification https://t.co/i9zTlAJ5rH
- Yang et al. (2026) Data Annotation with Large Language Models: Lessons from a Large Empirical Evaluation https://t.co/VGMv5mO7h5
- Heseltine (2026) Comparing Large Language Models for TextClassification: Model Selection Across Tasks,Texts, and Languages https://t.co/BGAxVO1Dak

[John B. Holbein @JohnHolbein1]

It's for this paper.

We'll release it soon.

[Robb Willer @RobbWiller]

🚨🚨New paper in @Nature with @AshuAshok, @lukebeehewitt & @ghezae_isaias 🚨🚨

Can LLMs predict results of social science experiments?

Across 70 preregistered studies, we find strong correspondence (r=.85) between LLM-predicted and observed effects.

https://t.co/hIrsjhrxRz 🧵

[American Political Science Review @apsrjournal]

From our FirstView: Beyond Belief Change: The Persuasive Returns of Targeting Attitude–Relevant Beliefs by @YamilRVelez, @patrickpliu and SCOTT CLIFFORD. https://t.co/qSwM9Qf465

[Sociology at Oxford @SociologyOxford]

🚨 New jobs! 🚨

We're looking for 2 Postdoctoral Research Fellows (1 quant, 1 qual), for a full-time, fixed-term position for up to 4 years!

The Fellows will have full control over the nature/direction of their research while here 📝

🔗https://t.co/RDACpSqKi0

[American Political Science Review @apsrjournal]

From our February Issue: Causal Panel Analysis under Parallel Trends: Lessons from a Large Reanalysis Study by ALBERT CHIU, XINGCHEN LAN, ZIYI LIU and YIQING XU. https://t.co/HU6lGfe467

[Eddie Yang @ey_985]

New paper in Nature. The more a government controls its domestic media, the more it dominates AI training data, the more pro-regime outputs we get from AI. By scraping the open web, LLMs are unwittingly laundering state-coordinated narratives into seemingly objective answers.

[scott cunningham @causalinf]

Results from my Claude Code audit of six Callaway and Sant'Anna packages (two in python, two in R, two in Stata). Same specification, same dataset, same covariates, same estimator, almost never do they agree.

https://t.co/KpRwlAagbr

[Yiqing Xu @xuyiqing]

🧵**Structural Conjoint**

New paper with Avidit Acharya and Jens Hainmueller:

“Learning Preferences from Conjoint Data: A Structural Deep Learning Approach.”

We show how to recover interpretable structural parameters from conjoint experiments, combining a random utility model with double-debiased machine learning (DML).

Paper: https://t.co/aygvFyizDT

[Patrick Liu @patrickpliu]

New paper w/ @YamilRVelez! A lot of great research on political microtargeting discounts personalization: tailored ads (using AI or not) rarely beat a single-best message. We define two types of microtargeting, clarify when tailoring matters, & showcase a novel audio-based design

[PA Journal @journal_pa]

Have you had a chance to read "Performance Rewards and Job Satisfaction in More and Less Developed Countries: Multi-Level Evidence From Bureaucrats in 10 Countries" in 104(1)? https://t.co/p20XC9vsLB

[Jack Santucci @jacksantucci]

The declarations of independents: Open‐ended survey responses and the nature of non‐identification - Allamong - American Journal of Political Science - Wiley Online Library https://t.co/iRjaOYMp1B

[Josh Kalla @j_kalla]

Which AI is most persuasive? New working paper w/ Zhongren Chen & Quan Le, we tested 7 frontier LLMs on 19k people. Ranking: (1) Claude; (2, tied) GPT, Gemini (3) Grok. Consistent across issues and bipartisan stances

[Haemin Jee @jee_haemin]

🧵 Thread: Does China actually respond to citizen complaints? Who's watching matters a lot. New paper in Governance 👇

[Tetsuro Kobayashi @tkobyashi]

The Sixth Wave data for the Asian Barometer Survey in Japan have now been released. We welcome applications for access to the survey data from interested individuals following the data release.
https://t.co/lCG0IeIS4k

[Eddie Yang @ey_985]

New paper at AJPS: "The limits of AI for authoritarian control." The more repression there is, the less information exists in AI's training data, and the worse the AI performs. Ironically, data from democracies can help improve repressive AI.

[Yiqing Xu @xuyiqing]

1/ 🐎 Our gift for the Year of the Horse:

An AI-assisted workflow that scales reproducibility in empirical research. w/ @YangYang_Leo

Paper: https://t.co/fslLN0zTQO

[Matt Grossmann @mattgrossmann]

AI's accuracy in censorship decreases with increasing repression, especially during times of political crisis
https://t.co/mHifgS6UH1

[Kiran Garimella @gvrkiran]

Great paper thinking about the role AI will play in entertainment, both good & bad.

We currently frame AI around intelligence and productivity, but one of the big use case is entertainment (likely the strategy behind Sora)

A shift from utility to culture
https://t.co/fK4QaqZAc9

[Hasan Toor @hasantoxr]

🚨BREAKING: Google just dropped another hit!

It's called PaperBanana and it generates publication-ready academic illustrations from just your methodology text.

No Figma. No manual design. No illustration skills needed.

Here's how it works:

A team of AI agents runs behind the scenes
→ One finds good diagram examples
→ One plans the structure
→ One styles the layout
→ One generates the image
→ One critiques and improves it

Here's the wildest part:

Random reference examples work nearly as well as perfectly matched ones. What matters is showing the model what good diagrams look like, not finding the topically perfect reference.

In blind evaluations, humans preferred PaperBanana outputs 75% of the time.

This is the recursion we've been waiting for AI systems that can fully document themselves visually.

Waitlist’s open, Link in the first comment.