About

Balaji Padmanabhan is the Associate Dean of Strategic Initiatives and a Professor of Decisions, Operations and Information Technologies at the Robert H. Smith School of Business. He holds a Bachelor's degree in Computer Science from the Indian Institute of Technology (IIT) Madras and a PhD from New York University’s Stern School of Business. With 25 years of experience, his work focuses on the design of artificial and augmented intelligence solutions that integrate data, machine learning, and modeling of real-world systems through complex systems simulations. His research has broad applications across various fields including business, policy, media, and healthcare. He has published extensively in data science and related areas in premier journals and conferences, and has served on the editorial boards of leading journals such as Management Science, MIS Quarterly, INFORMS Journal on Computing, Information Systems Research, Big Data, ACM Transactions on MIS, and the Journal of Business Analytics.

Research topics

• Computer Science
• Artificial Intelligence
• World Wide Web
• Data science
• Political Science
• Psychology
• Knowledge management
• Social psychology
• Engineering
• Statistics
• Mathematics
• Theoretical computer science

Selected publications

• What Exactly is a Deepfake?

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Information-Consistent Language Model Recommendations through Group Relative Policy Optimization

  ArXiv.org · 2025-12-14

  preprintOpen access

  Large Language Models (LLMs) are increasingly deployed in business-critical domains such as finance, education, healthcare, and customer support, where users expect consistent and reliable recommendations. Yet LLMs often exhibit variability when prompts are phrased with minor differences, even when semantically equivalent. Such inconsistency undermines trust, complicates compliance, and disrupts user experience. While personalization is desirable in certain contexts, many enterprise scenarios, such as HR onboarding, customer support, or policy disclosure, require invariant information delivery regardless of phrasing or prior conversational history. Existing approaches, including retrieval-augmented generation (RAG) and temperature tuning, improve factuality or reduce stochasticity, but cannot guarantee stability across equivalent prompts. In this paper, we propose a reinforcement learning framework based on Group Relative Policy Optimization (GRPO) to directly optimize for consistency. Unlike prior applications of GRPO, which have been limited to reasoning and code generation, we adapt GRPO to enforce the stability of information content across groups of semantically equivalent prompts. We introduce entropy-based helpfulness and stability rewards, treating prompt variants as groups and resetting conversational context to isolate phrasing effects. Experiments on investment and job recommendation tasks show that our GRPO-fine-tuned model reduces variability compared to the baseline LLM model. To our knowledge, this is a novel application of GRPO for aligning LLMs toward information consistency, reframing variability not as an acceptable feature of generative diversity, but as a correctable flaw in enterprise deployments.

  PublisherOA PDFDOI

• Do LLMs have a Gender (Entropy) Bias?

  arXiv (Cornell University) · 2025-05-24

  preprintOpen access

  We investigate the existence and persistence of a specific type of gender bias in some of the popular LLMs and contribute a new benchmark dataset, RealWorldQuestioning (released on HuggingFace ), developed from real-world questions across four key domains in business and health contexts: education, jobs, personal financial management, and general health. We define and study entropy bias, which we define as a discrepancy in the amount of information generated by an LLM in response to real questions users have asked. We tested this using four different LLMs and evaluated the generated responses both qualitatively and quantitatively by using ChatGPT-4o (as "LLM-as-judge"). Our analyses (metric-based comparisons and "LLM-as-judge" evaluation) suggest that there is no significant bias in LLM responses for men and women at a category level. However, at a finer granularity (the individual question level), there are substantial differences in LLM responses for men and women in the majority of cases, which "cancel" each other out often due to some responses being better for males and vice versa. This is still a concern since typical users of these tools often ask a specific question (only) as opposed to several varied ones in each of these common yet important areas of life. We suggest a simple debiasing approach that iteratively merges the responses for the two genders to produce a final result. Our approach demonstrates that a simple, prompt-based debiasing strategy can effectively debias LLM outputs, thus producing responses with higher information content than both gendered variants in 78% of the cases, and consistently achieving a balanced integration in the remaining cases.

  PublisherOA PDFDOI

• Code and Data Repository for Social Network Prediction Problems: Using Meta-Paths and Dynamic Heterogeneous Graph Representation for Label Propagation

  INFORMS journal on computing · 2025-11-10

  article

  This work introduces a dynamic heterogeneous graph representation that integrates time into both nodes and edges, enabling a more accurate modeling of multiplex and evolving relationships in social networks. We further propose Meta-paths + LPA, an extension of the Label Propagation Algorithm that incorporates temporal meta-paths for improved classification on dynamic heterogeneous structures. The framework is demonstrated through a case study on Steemit, showcasing its ability to handle complex time-dependent queries and prediction tasks.

  PublisherDOI

• What are patients watching online? Using recommender systems and large language models to discover temporal viewership patterns in maternal health

  Expert Systems with Applications · 2025-06-09

  article
  PublisherDOI

• Deepfakes for Good: Empirical Analysis and AI Agentic Framework for Bias Measurement and Mitigation

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• What Exactly is a Deepfake?

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Social Network Prediction Problems: Using Meta-Paths and Dynamic Heterogeneous Graph Representation for Label Propagation

  INFORMS journal on computing · 2025-11-10

  article

  Graph representations for real-world social networks in the past have missed two important elements: (i) the multiplexity of connections and (ii) representing time. This paper presents a dynamic heterogeneous graph representation for social networks that includes time in every component of the graph, that is, nodes and edges, each of different types that captures heterogeneity. We illustrate the power of this representation by presenting four time-dependent queries and a multiclass classification problem that cannot easily be handled in conventional homogeneous graph representations. As a proof of concept, we present a detailed representation of a relatively new social media platform (Steemit), which we use to illustrate both the dynamic querying capability, as well as a prediction task using label propagation algorithm (LPA). We also present temporal social media meta-paths to generalize the LPA to dynamic heterogeneous graph structures, that is, Meta-paths + LPA. To validate and compare our proposed method, we conduct an experiment using three benchmark data sets and show that our proposed method outperforms almost all four state-of-the-art algorithms in category prediction task by at least 13.79% accuracy. History: Accepted by Ram Ramesh, Area Editor for Data Science & Machine Learning. Supplemental Material: The software that supports the findings of this study is available within the paper and its Supplemental Information ( https://pubsonline.informs.org/doi/suppl/10.1287/ijoc.2023.0274 ) as well as from the IJOC GitHub software repository ( https://github.com/INFORMSJoC/2023.0274 ). The complete IJOC Software and Data Repository is available at https://informsjoc.github.io/ .

  PublisherDOI

• Moving From ‘Hallucinations’ to ‘Fabrications’

  Psychiatric News · 2025-02-25

  article
  PublisherDOI

• From Deception to Perception: The Surprising Benefits of Deepfakes for Detecting, Measuring, and Mitigating Bias

  arXiv (Cornell University) · 2025-02-16

  preprintOpen access

  While deepfake technologies have predominantly been criticized for potential misuse, our study demonstrates their significant potential as tools for detecting, measuring, and mitigating biases in key societal domains. By employing deepfake technology to generate controlled facial images, we extend the scope of traditional correspondence studies beyond mere textual manipulations. This enhancement is crucial in scenarios such as pain assessments, where subjective biases triggered by sensitive features in facial images can profoundly affect outcomes. Our results reveal that deepfakes not only maintain the effectiveness of correspondence studies but also introduce groundbreaking advancements in bias measurement and correction techniques. This study emphasizes the constructive role of deepfake technologies as essential tools for advancing societal equity and fairness.

  PublisherOA PDFDOI

Frequent coauthors

• Lina Bouayad

  31 shared

• Arash Barfar

  University of Nevada, Reno

  14 shared

• Anna Ialynytchev

  Rehabilitation Research and Development Service

  14 shared

• Alexander Tuzhilin

  New York University

  13 shared

• Alan R. Hevner

  University of South Florida

  11 shared

• Kaushal Chari

  10 shared

• Negar Maleki

  University of South Florida

  8 shared

• Zhiqiang Zheng

  The University of Texas at Dallas

  8 shared

Awards & honors

• Smith’s Center for AI in Business Awarded $1M+ Federal Grant…