Hi, I’m Shivani Kumar

I’m a Postdoctoral Research Fellow at the BlaBlaBlab at University of Michigan, where I am working with Prof David Jurgens on culturally enriched and morally refined language models.

I completed my PhD from the LCS2 Lab, advised by Prof Tanmoy Chakraborty, from IIITD, before which I had completed my masters (M.Sc.) and bachelors (B.Sc.) from University of Delhi. My PhD journey took me deep into the territory of conversational AI, where I explored the lucidity of dialogue - exploring everything from emotions, humour, and sarcasm to the unique essence of each speaker.

My research interests lies into exploring how Large Language Models tackle rhetorical analysis, especially in conversations. I am interested in probing the three verticals of rhetoric: ethos, pathos, and logos. For ethos, I’m excited to examine into how cultural values, credibility, and trust shape dialogue. With pathos, I’m keen on understanding how emotions are managed and expressed in LLM interactions. And for logos, I wish to analyze how these models craft and navigate logical arguments.