David Jurgens

Identity-Aware AI Workshop announcement. Co-located with ECAI 2025 in Bologna on October 25, with submission deadline August 22. Topics include: Methods for effective, fair, and inclusive AI; Critiques of AI on the exclusion of identities; Methods for detecting and controlling bias; Perspectivist approaches to AI. Submission types: Long papers (8 pages), Short papers (4 pages), Extended abstracts, Mixed-media submissions (videos, blogs, codebase, artworks). For details, visit: identity-aware-ai.github.io

Wondering what makes each of us unique and how AI should handle human diversity? 🤔

We're organizing Identity-Aware AI workshop at #ECAI2025 Bologna on Oct 25.

Deadline: Aug 22
Website: identity-aware-ai.github.io

Best Resource Paper (1/2) #ACL2025NLP

Feel free to reach out on BlueSky or email if you want to connect (I'm terrible at checking Whova)

I'll be at #ACL2025 with a variety of papers on podcasts, linguistic accommodation, whether papers cite accurately, multilingual morals, demographically-prompted LLMs, how tokenization affects style recognition, and new multilingual data for author attribution. Looking forward to catching up!

Michigan allows it and I haven't heard of any issues from the sponsored research office about the terms

How Scientists Use Large Language Models to Program | Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems You will be notified whenever a record that you have chosen has been cited.

She also had a cool recent paper on this topic too dl.acm.org/doi/full/10....

There's also an opt-in raffle for five $100 cash awards. Scientists from any research area, career track, or stage (including student research assistants) are invited. If you want to share your experiences umich.qualtrics.com/jfe/form/SV_...

Do you using GenAI to help code? My colleague Elle O'Brien is running a study to answer questions about your programming practices and tool usage, including experiences with AI code assistants. The survey is anonymous and should take no more than 10-15 minutes to complete.

For Prospective PhD Students I’m recruiting PhD students who will begin their degree in Fall 2026! I am an incoming assistant professor at Wisconsin-Madison’s Computer Sciences department, and my research focuses on natural langu...

I'm sadly not at #IC2S2 😭, but I will be at #ACL2025 in Vienna ☕️ next week!!

Please spread the word that I'm recruiting prospective PhD students: lucy3.notion.site/for-prospect...

Open PhD positions in Denmark! daracademy.dk/fellowship/f...

If you want to apply to work with me and Johannes Bjerva at @aau.dk Copenhagen, I'll be at @ic2s2.bsky.social this week and @aclmeeting.bsky.social next week! DM me if you'd like to meet :)

There are a bunch of other interesting experiments in the paper too. I'll be around ACL too so come chat with any of us

This isn't to say that some future model can't learn these preferences BUT it strongly suggests that we need to be more careful when when trying to model different sources of systematic rating variance. If we hadn't also tested annotator variance we would have incorrectly thought demographics helped

... it turns out LLMs are mostly treating demographic combinations as annotators' IDs and just learning individual preferences. These are useful too and just leaning annotator preferences does equally as well as using demographics. But adding demographics with annotator IDs didn't help at all!

This paper was such a fun collaboration—and a surprising result. I fully expected we could train LLMs to learn demographic preferences in ratings but

Wanna do some authorship attribution? Chances are what tokenizer you use matters.

Tokenization is Sensitive to Language Variation, probably, more investigation necessary...

📄 ACL Findings paper: arxiv.org/pdf/2502.15343
🧑‍🏫 @dongng.bsky.social @davidjurgens.bsky.social and myself

See you at ACL!

I'm looking for two emergency reviewers for ARR for papers on agentic LLMs and on LLM interpretability. If you could get a review done in the next 1-2 days and would be open to helping, please DM me.

Our critical perspective of AI efficiency is now online in Communications of the ACM!

cacm.acm.org/sustainabili...

Despite the duality of these results, I personally have a lot of hope for this tech. With a more personalized and more sophisticated design for how GenAI assists, I think we can mitigate the negative perceptions while still making it easier for people to authentically participate.

GenAI generally helps people engage in conversations they wouldn't otherwise and can help bring in a more diverse group of conversants. However, readers dislike GenAI-assisted content more (👎) and rate the content quality lower.

There are so many cool results in the paper, so I encourage you to check out Ander's *interactive* webpage for the paper (🔥🔥🔥) and/or read the paper itself. My quick summary of the main results...

Some types of GenAI assistance can improve authors' experiences and also improve readers' experiences—but many types of assistance that help one group end up making the experience worse for the other. As Anders notes, the duality of GenAI!

Using real people having real conversations, we tested how different kinds of GenAI support influenced their behavior as authors and how they felt as readers in those conversations—and compared these to human-only conversations on the same topics. Short summary...

ChatGPT is writing this quote post. Actually, not. But if it did, would it write a better post than I could as an author—and would you as a reader be more likely to engage with it? We wanted to test this and the fearless @andersgiovanni.com built a whole social media platform for us to experiment...

⏰ Poster submission deadline TODAY for the pre-@aclmeeting.bsky.social Workshop 🇩🇰
📅 We've finalised our schedule and speakers: Thamar Solorio, Smaranda Muresan, @davidjurgens.bsky.social, Nanyun Peng, @danishpruthi.bsky.social , Kai-Wei Chang, and Mausam www.aicentre.dk/events/pre-a... @aicentre.dk

Today, NSF added another 500 GRFP awardees, bringing this year’s total to 1500
www.research.gov/grfp/Awardee...

Write with Science Homecoming, and put your piece on the map:

sciencehomecoming.com

Do folks really find the ARR checklist valuable enough to justify that a paper submission takes this much effort?

Congrats!! UW is great and Madison is a fantastic city. Welcome to the midwest 😎

The faces of the authors for the NAACL 2025 paper Causally Modeling the Linguistic and Social Factors that Predict Email Response

This paper was also a group effort by the whole lab as a part of a "research jam" to learn from each other by collaborating on a single project. It was quite the dream team of folks and a fun experience.

Causally Modeling the Linguistic and Social Factors that Predict Email Response Yinuo Xu, Hong Chen, Sushrita Rakshit, Aparna Ananthasubramaniam, Omkar Yadav, Mingqian Zheng, Michael Jiang, Lechen Zhang, Bowen Yi, Kenan Alkiek, Abraham Israeli, Bangzhao Shu, Hua Shen, Jiaxin Pei,...

If you want the full scoop, come talk to Nancy Xu at our poster Thursday 4-5:30 in the CSS area to find out more! @bowenyi.bsky.social is also around #NAACL too if you see him. And check out the paper for all the details aclanthology.org/2025.naacl-l...