Meet Dr Procheta Sen

Dr Procheta Sen, Lecturer in Computer, is part of the Natural Language Processing group and focuses on making AI language models more transparent, fair, and accessible to all.

• Name – Dr Procheta Sen
• Position – Lecturer in Computer Science
• Group Name – Natural Language Processing, Department of Computer Science
• Joined University of Liverpool – 2022
• Born – West Bengal, India
• PhD – Dublin City University, Ireland

What is your research about?

I broadly work at the intersection of natural language processing and machine learning. Recently my focus has been on explainability which aims to make the internal mechanisms of an AI model transparent and interpretable. The goal is not only to show what the model predicted, but why it arrived at that prediction and when it might fail. The end user of an AI model with explainability can be a domain expert, such as clinicians, law experts, a lay person who doesn’t have any knowledge about AI models, or an AI practitioner. We focus on developing explainability solutions broadly across three different categories:

a) Post-hoc (explanations after training): feature attribution, counterfactuals, example-based explanations, and surrogate models to summarise behaviour.

b) Mechanistic interpretability: opening the “black box” to map computations inside neural networks - identifying neurons, circuits, and pathways that implement specific functions

c) Intrinsic (interpretable by design): models or components with human-readable structure (rules, graphs).

What or who first inspired you to be interested in your research subject?

I have always loved how language shapes how we think and connect. Seeing how machine learning can model and amplify that power convinced me to pursue natural language processing because it blends seamlessly with any field and can drive real-world impact.

What are you most proud of achieving during your research career so far?

I am proud of my work on debiasing and explainability. On the explainability side, I have developed models that make outputs understandable to both AI practitioners and lay people.

What techniques and equipment do you use to conduct your research?

Broadly, I apply various machine learning methodologies to study models’ inner workings and the reasons behind their behaviour.

What is the key to running a successful research group?

I think a successful research group hinges on two things: identifying talent and fostering it with compassion.

What impact is your research having outside of academia?

Most of our work is open source, and several of our methods have been deployed in real-world settings, such as the legal sector. I also organise the NLP for Social Good symposium every year which brings together researchers from multiple disciplines to explore responsible applications of natural language processing and AI.

How do you plan to develop your research in the future?

I plan to advance my research by building intelligent systems that are responsible by design. Being responsible spans multiple dimensions. I would like to particularly focus on fairness, transparency, and equity.

What problem would you like to solve in the next 10 years through your research?

I aim to develop methods that make language models more transparent and equitable, so their benefits are accessible to everyone.

What advice would you give to someone considering a career in research?

Focus on the pillars: probability and statistics, linear algebra, and strong programming, these will carry you in any research area.

Where can readers learn more about your research?

I share details about my exciting research projects on my website: https://procheta.github.io/