I’m actively seeking Ph.D. positions starting in Fall 2027 in Responsible AI.

Hi! I’m Pavithra 🌱

I study when and why large language models treat people differently, and what those differences mean for the people who rely on them. My work audits LLMs for demographic, linguistic, and cultural bias across various settings including economic advice and moral judgments. I’m also interested in how AI assistance shapes human decision-making.

Currently, I work as a scientist at Amrita Vishwa Vidyapeetham, where I am advised by Prof. Krishnashree Achuthan. Previously, I worked as a researcher at TCS Research, advised by Dr. Preethu Rose Anish.

Here’s what I’ve been pondering lately:

• 🌱 When do differences in an LLM’s responses reflect legitimate cultural variation, and when are they unfair treatment?
• 🌱 How can we discover biases we never thought to test for?
• 🌱 How do biases evolve in naturalistic, multilingual conversations, and how do they shape the decisions users make?
• 🌱 How can we make the factors behind a model’s response transparent to users, so they can question the advice they get?