Koustuv Saha

Excited about my new op-ed on the observer effect in the age of LLMs and Generative AI published at @techpolicypress.bsky.social. Thanks for the amazing mentorship, @micheleweldon.bsky.social .

The Risks of the 'Observer Effect' from Being Watched by AI | TechPolicy.Press Dr. Koustuv Saha says if users begin to feel their privacy is compromised, they may stop using AI in the very ways that make it useful.

One reason LLMs are so popular is the promise of seemingly private, judgment-free interaction. But once users internalize that their data is being recorded, studied, or reused, it could lead to a shift in how people think, speak, and even feel when using AI, says computer scientist Koustuv Saha.

When one LLM can’t handle all the chaos of online moderation, assemble a team!
Agam Goyal's #EMNLP2025 paper ft. @kous2v.bsky.social, @ceshwar.bsky.social introduces MoMoE, a Mixture of Moderation Experts where models work toward fairer, scalable moderation.
📝https://go.illinois.edu/emnlp2

Very proud of @melissashi.bsky.social and @keranwang.bsky.social for leading our #CSCW2025 paper exploring the mental health needs of AD/ADRD family caregivers---a collaboration with @dwyoo.bsky.social and @ravikarkar.bsky.social

dl.acm.org/doi/10.1145/...

Thanks a lot, Alexandra. :)

Thank you so much, Emre! ❤️

Thanks so much, dada! ❤️

I'll never forget the moment my first #ICWSM paper was accepted in my PhD's 2nd year. I told @munmun10.bsky.social, "I can't explain how happy I am." I was overjoyed to be a part of this community. Each ICWSM has been so enlightening and emotional---and this recognition brought it all back. ❤️

So honored to receive this recognition. Can't thank enough those who supported me along the way. Feels like a dream to receive an award named after two legends, @ladamic.bsky.social @nglance.bsky.social. #ICWSM is special to my heart, sad to have missed it this year. Thank you, @icwsm.bsky.social. ❤️

Our recent work at JMIR Aging, with @ravikarkar.bsky.social, and led by Sidharth, Chunyu, and Yoshee.

In another #CHI2025 paper led led by Charlotte Lambert, in collaboration with @kous2v.bsky.social, we conducted a quasi-experimental study and found that Reddit users who received “high scores” and “gold” made more frequent and higher quality posts (i.e., higher scores and fewer removals).

Continuing our work on LLMs for content moderation, our #NAACL2025 shows that SLMs can outperform LLMs with higher accuracy, better recall, and easier fine-tuning for community norms—making them a scalable, convenient, and cost-effective choice. /w @agam-goyal.bsky.social @ceshwar.bsky.social

Q: Are fine-tuned SLMs better than LLMs at identifying online content that violate community norms?🤖⚖️

A: Yes!

Find out more in our #NAACL2025 paper co-lead by @agam-goyal.bsky.social and Xianyang Zhan, in collaboration with Yilun Chen and @kous2v.bsky.social.

📄 arxiv.org/pdf/2410.13155

CHI 2026 Open Call for Volunteers Take this survey powered by surveymonkey.com. Create your own surveys for free.

CHI 2026 Barcelona Open Call for volunteers is Open! Please sign up to join the team! This is for organizing and technical committees (excluding subcommittee chairs and area chairs for the papers program…that comes later). Details in the submission form. www.surveymonkey.com/r/GGPYYPD

A photo of Boulder, Colorado, shot from above the university campus and looking toward the Flatirons.

I'm recruiting 1-2 PhD students to work with me at the University of Colorado Boulder! Looking for creative students with interests in #NLP and #CulturalAnalytics.

Boulder is a lovely college town 30 minutes from Denver and 1 hour from Rocky Mountain National Park 😎

Apply by December 15th!