Bridge-AI Lab

We are thrilled to announce that Yash Mahajan has successfully passed his PhD defense as the 5th PhD graduate from the BridgeAI Lab. Please join us in celebrating his achievement and wishing him the very best as he steps into the next exciting chapter of his career.

Congratulations, Dr. Yash! 🎓✨

Do large language models think like humans? Are they also prone to human-like cognitive biases?

We just launched a new ranking system of LLMs based on their ability to resist cognitive biases with a large-scale study of 2.8M+ responses across 8 well-known biases (Anchoring, Availability, Confirmation, Framing, Prospect Theory & more).

See which models resist bias the best, how prompt design changes outcomes, and why this matters for trustworthy decision making with AI.

Read our ArXiv paper detailing the experiments and results here: https://arxiv.org/abs/2509.22856

Explore the live rankings here: https://bridgeai-lab.github.io/LLM-Ranking/

LLM Rankings Cognitive Bias Ranking evaluates and compares Large Language Models across key cognitive bias categories to assess robustness in reasoning.

Just released our latest survey, "Multi-Party Conversational Agents: A Survey", on arXiv!

We delve into the complexities of Multi-Party Conversational Agents (MPCAs), exploring three themes:
State of Mind Modeling: Understanding participants' mental statesSemantic Understanding: Grasping the nuances of dialogue contentAgent Action Modeling: Predicting future conversation flows
Our analysis spans from classical machine learning approaches to the latest in Large Language Models (LLMs) and multi-modal systems. We identify and highlight the importance of Theory of Mind (ToM) and multi-modality in building intelligent MPCAs.

Read the full paper here: https://arxiv.org/abs/2505.18845

Multi-Party Conversational Agents: A Survey Multi-party Conversational Agents (MPCAs) are systems designed to engage in dialogue with more than two participants simultaneously. Unlike traditional two-party agents, designing MPCAs faces additional challenges due to the need to interpret both utterance semantics and social dynamics. This survey...

Bridge-AI Lab is looking for a talented Postdoctoral Researcher to advance cutting-edge AI/ML research! If you have a Ph.D. and more than one publication at top AI/ML conferences or journals, we want to hear from you!
🔹 Minimum Requirements:
- Ph.D. in Machine Learning or related field
- At least two publications in reputed AI/ML conferences or journals

🔹 Preferred Qualification:
- Experience in grant proposal writing

Salary: $60K + Benefits
Join us in pushing the boundaries of AI research! If you're interested, apply now or share with potential candidates.

For inquiries, contact: [[email protected]]

We are glad to share our latest NAACL 2025 paper, "LLMs as Meta-Reviewers' Assistants: A Case Study".

This paper investigates how LLMs can assist meta-reviewers by generating a controlled multi-perspective summary (MPS) of peer-reviewers opinions. Through a systematic qualitative study, the paper analyzes how prompting LLMs with varying levels of details can impact the quality of MPS generated by them.

Read the full paper here:

arxiv.org

We are pleased to announce that the final version of our paper, "ALIGN-SIM: A Task-Free Test Bed for Evaluating and Interpreting Sentence Embeddings through Semantic Similarity Alignment," has been published at EMNLP 2024!

A big thanks to Dr. Eduardo Blanco for his collaboration and significant contributions to this work.

Our paper introduces ALIGN-SIM, a novel task-free framework for evaluating and interpreting sentence embeddings based on various semantic similarity alignment criteria.

Read the full paper here:

ALIGN-SIM: A Task-Free Test Bed for Evaluating and Interpreting Sentence Embeddings through Semantic Similarity Alignment Yash Mahajan, Naman Bansal, Eduardo Blanco, Santu Karmaker. Findings of the Association for Computational Linguistics: EMNLP 2024. 2024.

We are excited to collaborate with to introduce BongLLaMA, the first ever and, of course, state-of-the-art Bangla Open Source LLM (Based on Meta LLaMA). Try it out.

University of Central Florida

BongLLaMA: LLaMA for Bangla Language Bangla (or "Bengali") is a language spoken by approximately 240 million native speakers and around 300 million people worldwide. Despite being the 5th largest spoken language in the world, Bangla is still a "low-resource" language, and existing pretrained language models often struggle to perform we...

We are excited to share that Dr. Souvika Sarkar, who recently completed her Ph.D. with the BDI Lab under the supervision of Dr. Santu, has joined Wichita State University as a tenure-track assistant professor for the School of Computing. Dr. Sarkar is actively seeking motivated Ph.D. students to join her research team, which will focus on exploring the potential of generative AI models to develop innovative, real-world solutions. Interested candidates are encouraged to email their CVs to Dr. Sarkar at [email protected]. For more information, please visit her homepage at https://www.wichita.edu/profiles/academics/engineering/SoC/01_Faculty/Sarkar-Souvika.php

We extend our best wishes to Dr. Sarkar as she embarks on this exciting new chapter in her career.