Preetam Dammu

PhD student at @UW | ML research

[ACM CHIIR 2026 @ACM_CHIIR]

@UW @hideo_joho @preetams7 Great to see @preetams7 and Tanya Roosta leading an engaging session on Information Seeking in the Age of Agentic AI. 🤖💡

[Preetam Dammu @preetams7]

(2/2)
🔗 Tutorial details:
https://t.co/YeAuvHvAab
📝 Conference registration:
https://t.co/QGQs3srxi5

If you’re interested in IR, HCI, agentic AI, or the future of search and decision support, we’d love to have you join us.

#IR #AgenticAI @ACM_CHIIR @infoseeking1

[Preetam Dammu @preetams7]

(1/2)
🚨 CHIIR 2026 registration is now open!

Our tutorial, Information Seeking in the Age of Agentic AI, examines how AI systems are shifting from supporting search to actively doing the seeking on users’ behalf, and what this means for design and evaluation in practice.

[Preetam Dammu @preetams7]

Dynamic-KGQA tackles this by dynamically generating test samples, reducing leakage & making evaluations more robust and closer to real-world performance.

More details here:
🔗 https://t.co/XvyctopMXC
📄 https://t.co/92E9InRpWa

📍 Giotto Hall | 🕑 2 PM | @SIGIRConf

[Preetam Dammu @preetams7]

Excited to present at #SIGIR2025!

Dynamic-KGQA: Framework for Generating Adaptive QA Datasets
(@HimanshuJNaidu, @chirag_shah)

With generative models trained on anything & everything, static datasets often leak into pretraining — blurring the line between capabilities & memory.

[Preetam Dammu @preetams7]

2️⃣ Unveiling Covert Harms in LLM Conversations
📍 Brickell | Nov 14 | 10:30-12 PM
🔗 https://t.co/WFDDLNZ9rb

(W/ @hayounggjung, @anjali_singh35, @monojitchou, @tanmit)

Excited to explore these topics together – please join us at our sessions!

#EMNLP2024 #EMNLP
(2/2)

[Preetam Dammu @preetams7]

Excited to present two papers at #EMNLP2024!

1️⃣ ClaimVer: Explainable Claim Verification using Knowledge Graphs
📍 Session 12 | Nov 14 | 2-3:30 PM
🔗 https://t.co/FVPrCM83Y1

(W/ @i_amanchadha, @chirag_shah)

(1/2)

[Dimitris Papailiopoulos @DimitrisPapail]

I guess we need a lot of things

[Omar Alonso @elunca]

Our team has a couple of intern papers at SIGIR 2024. Great work by Preetam Dammu (UW) and Diji Yang (UCSC). @AmazonScience

Answering scoring framework for LLMs

https://t.co/SE9zENU0C2

Question intent taxonomy https://t.co/wkx3LA0HiO

[Monojit Choudhury @monojitchou]

Despite much noise and work on #alignment, it is amazing how brittle still #LLMs are. It doesn't take much effort to bring out their ugly biases if one probes beyond popular topics of gender, race and religion.
Our new study explores #caste biases in job interview settings.

[Preetam Dammu @preetams7]

5/ Our findings raise concerns regarding the suitability of LLMs in sensitive applications, such as recruitment, conversation generation, and screenplay-writing/script generation, which may generate texts containing CHAST. Preprint: https://t.co/2axTFst9Fl

[Preetam Dammu @preetams7]

LLMs are known to generate harmful views, but what are the various and potentially covert forms of harm and identity threats in LLM-generated conversations?

We explore this with @hayounggjung, @anjali_singh35, @monojitchou, and @tanmit. Preprint: https://t.co/2axTFst9Fl

[Preetam Dammu @preetams7]

4/ We found that multiple LLMs generated malign views in seemingly neutral language, unlikely to be detected and flagged by SOTA tools such as PerspectiveAPI. Our method, by contrast, is specifically designed to capture these covert harms.

[infoseeking @infoseeking1]

Excited to present @preetams7 latest work with @yunhefengit @chirag_shah at #IJCAI2023. Paper link: https://t.co/7w58pVv5vM #GenerativeAI #MachineLearning #EthicsInTech #Inclusion #AI #MachineLearningBias #WebSearch #GenerativeModels