Hi! It's me, Sushrita Rakshit 👋

It's me, Sushrita Rakshit!

I'm a recent graduate of the University of Michigan, where I studied computer science and statistics and worked with Dr. David Jurgens. My research focuses on modeling human social behavior from language and using these insights to design more aligned and socially aware AI systems. I’ve worked on emotionally and culturally grounded dialogue models, large-scale causal analyses of human communication patterns, and evaluations of how well language models simulate human behavior. My undergraduate research was recognized with a CRA Outstanding Undergraduate Researcher Honorable Mention.

PhD Applications: I am applying for PhD positions this cycle (Fall 2026 start)! Please reach out to me if you see any compatibility in our research interests!
NUTMEG is an alternative to MACE for identifying ground truth when groups of annotators systematically disagree.

Towards Emotionally-Aware Agents for Dispute Resolution

Sushrita Rakshit, James Hale, Kushal Chawla, Jeanne Brett, Jonathan Gratch
Affective Computing and Intelligent Interaction (ACII)

KODIS: A Multicultural Dispute Resolution Dialogue Corpus

James Hale, Sushrita Rakshit, Kushal Chawla, Jeanne M. Brett, Jonathan Gratch
North American Chapter of the Association for Computational Linguistics (NAACL)

Please Reply: 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, Haotian Zhang, Miriam Schirmer, David Jurgens
North American Chapter of the Association for Computational Linguistics (NAACL)
You can get systems to learn individual and group behaviors during annotation but only if you add structural priors to the model

Towards bidirectional human-AI alignment: a systematic review for clarifications, framework, and future directions. vol. abs/2406.09264

Hua Shen, Tiffany Knearem, Reshmi Ghosh, Kenan Alkiek, Kundan Krishna, Yachuan Liu, Ziqiao Ma, Savvas Petridis, Yi-Hao Peng, Li Qiwei, Sushrita Rakshit, Chenglei Si, Yutong Xie, Jeffrey P. Bigham, Frank Bentley, Joyce Chai, Zachary Lipton, Qiaozhu Mei, Rada Mihalcea, Michael Terry, Diyi Yang, Meredith Ringel Morris, Paul Resnick, David Jurgens
NeurIPS 2025 Position Track
Optimizing the system and task parts of the prompt can have huge benefits

Real or robotic? Assessing whether LLMs accurately simulate qualities of human responses in dialogue

Jonathan Ivey, Shivani Kumar, Jiayu Liu, Hua Shen, Sushrita Rakshit, Rohan Raju, Haotian Zhang, Aparna Ananthasubramaniam, Junghwan Kim, Bowen Yi, Dustin Wright, Abraham Israeli, Anders Giovanni Møller, Lechen Zhang, David Jurgens
preprint.

In My Undergraduate Days:

Fall 2024: SI 301 -- Models of Social Information Processing (Network Science and Intro Game Theory)

Winter 2025: SI 339 -- Web Design, Development, and Accessibility

Other Cool Things I Teach/Run:

Indian Classical Music: Vocals (Since forever - Present)

Knowledge Cafe @Verizon: Weekly AI seminars with hands-on demos for people interested in growing and sharing their work

Broadly, I conduct research in the areas of natural language processing, human-ai collaboration, and some computational social science. The best way to define my work, as quoted by Hua Shen, is to "maximize compatibility and minimize risk" in Human-AI Co-evolution. Below are a few themes I break my work into:

  • Human-AI Alignment and Socially Intelligent NLP: I design and evaluate language technologies that align with human values, behavior, and expectations—focusing on how humans and AI systems can understand, adapt to, and collaborate with one another in communication.
  • Language Models as Social Measurement Proxies: I use language models to uncover latent patterns in human communication, analyzing multi-turn dialogue, emotional trajectories, and social networks.
  • Emotionally and Culturally Grounded Dialogue Systems: My work develops conversational agents that are sensitive to emotion, culture, and context—especially in high-stakes, moral, and emotional settings—by combining NLP with insights from psychology, social dynamics, and cross-cultural communication.

My long term goal encompasses build AI systems that can genuinely understand and communicate with people, enabling more trustworthy, collaborative, and equitable human-AI interaction. I mean this as not just through language, but through shared social, emotional, and cultural understanding

.

Q: How do I pronounce your last name?
A: Roh-kheeth

Q: Why the mismatch?
A: 🤷

Q: Ok...what about your first name?
A: Sue-shree-tah

Q: What are you interested in?
A: Running, reading, and cooking. I try and make a new eggless recipe every week (thanks to my allergic roomie). I'm also addicted to using Strava and tracking my mile per hour records. In the Florida heat, it's an impressive brag.