Monojit Choudhury
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Monojit Choudhury has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Sougata Saha
University of Artificial Intelligence
4 shared publications Aditi Khandelwal
Microsoft Corporation
3 shared publications- Kumar Tanmay
Microsoft Corporation
3 shared publications - Saurabh Kumar Pandey
MBZUAI
3 shared publications - Anjali Singh
University of Washington
2 shared publications - Atharva Mehta
University of Artificial Intelligence
2 shared publications
Latest Publications
- Open Access
To Generate or Discriminate? Methodological Considerations for Measuring Cultural Alignment in LLMs
Socio-demographic prompting (SDP) - prompting Large Language Models (LLMs) using demographic proxies to generate culturally aligned outputs - often shows LLM responses as stereotypical and biased. While effective in assessing LLMs' cultural competency, SDP is prone to confounding factors such as prompt sensitivity, decoding parameters, and the inherent difficulty of generation over discrimination tasks due to larger output spaces. These factors complicate interpretation, making it difficult to determine if the poor performance is due to bias or the task design. To address this, we use inverse socio-demographic prompting (ISDP), where we prompt LLMs to discriminate and predict the demographic proxy from actual and simulated user behavior from different users. We use the Goodreads-CSI dataset (Saha et al., 2025), which captures difficulty in understanding English book reviews for users from India, Mexico, and the USA, and test four LLMs: Aya-23, Gemma-2, GPT-4o, and LLaMA-3.1 with ISDP. Results show that models perform better with actual behaviors than simulated ones, contrary to what SDP suggests. However, performance with both behavior types diminishes and becomes nearly equal at the individual level, indicating limits to personalization.
Open AccessTo Generate or Discriminate? Methodological Considerations for Measuring Cultural Alignment in LLMs
Socio-demographic prompting (SDP) - prompting Large Language Models (LLMs) using demographic proxies to generate culturally aligned outputs - often shows LLM responses as stereotypical and biased. While effective in assessing LLMs' cultural competency, SDP is prone to confounding factors such as prompt sensitivity, decoding parameters, and the inherent difficulty of generation over discrimination tasks due to larger output spaces. These factors complicate interpretation, making it difficult to determine if the poor performance is due to bias or the task design. To address this, we use inverse socio-demographic prompting (ISDP), where we prompt LLMs to discriminate and predict the demographic proxy from actual and simulated user behavior from different users. We use the Goodreads-CSI dataset (Saha et al., 2025), which captures difficulty in understanding English book reviews for users from India, Mexico, and the USA, and test four LLMs: Aya-23, Gemma-2, GPT-4o, and LLaMA-3.1 with ISDP. Results show that models perform better with actual behaviors than simulated ones, contrary to what SDP suggests. However, performance with both behavior types diminishes and becomes nearly equal at the individual level, indicating limits to personalization.
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2319-2330
How Deep Is Representational Bias in LLMs? The Cases of Caste and Religion
Representational bias in large language models (LLMs) has predominantly been measured through single-response interactions and has focused on Global North-centric identities like race and gender. We expand on that research by conducting a systematic audit of GPT-4 Turbo to reveal how deeply encoded representational biases are and how they extend to less-explored dimensions of identity. We prompt GPT-4 Turbo to generate over 7,200 stories about significant life events (such as weddings) in India, using prompts designed to encourage diversity to varying extents. Comparing the diversity of religious and caste representation in the outputs against the actual population distribution in India as recorded in census data, we quantify the presence and “stickiness” of representational bias in the LLM for religion and caste. We find that GPT-4 responses consistently overrepresent culturally dominant groups far beyond their statistical representation, despite prompts intended to encourage representational diversity. Our findings also suggest that representational bias in LLMs has a winner-takes-all quality that is more biased than the likely distribution bias in their training data, and repeated prompt-based nudges have limited and inconsistent efficacy in dislodging these biases. These results suggest that diversifying training data alone may not be sufficient to correct LLM bias, highlighting the need for more fundamental changes in model development.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (3) , 2319-2330
Open AccessHow Deep Is Representational Bias in LLMs? The Cases of Caste and Religion
Representational bias in large language models (LLMs) has predominantly been measured through single-response interactions and has focused on Global North-centric identities like race and gender. We expand on that research by conducting a systematic audit of GPT-4 Turbo to reveal how deeply encoded representational biases are and how they extend to less-explored dimensions of identity. We prompt GPT-4 Turbo to generate over 7,200 stories about significant life events (such as weddings) in India, using prompts designed to encourage diversity to varying extents. Comparing the diversity of religious and caste representation in the outputs against the actual population distribution in India as recorded in census data, we quantify the presence and “stickiness” of representational bias in the LLM for religion and caste. We find that GPT-4 responses consistently overrepresent culturally dominant groups far beyond their statistical representation, despite prompts intended to encourage representational diversity. Our findings also suggest that representational bias in LLMs has a winner-takes-all quality that is more biased than the likely distribution bias in their training data, and repeated prompt-based nudges have limited and inconsistent efficacy in dislodging these biases. These results suggest that diversifying training data alone may not be sufficient to correct LLM bias, highlighting the need for more fundamental changes in model development.
- Open Access
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (1) , 217-226
Sacred or Synthetic? Evaluating LLM Reliability and Abstention for Religious Questions
Despite the increasing usage of Large Language Models (LLMs) in answering in questions in a variety of domains, their reliability and accuracy remain unexamined for a a plethora of domains, including the religious domains. In this paper, we introduce a novel benchmark FiqhQA focused on the LLM generated Islamic rulings explicitly categorized by the four major Sunni schools of thought, in both Arabic and English. Unlike prior work, which either overlooks the distinctions between religious schools of thought or fails to evaluate abstention behavior, we assess LLMs not only on their accuracy but also on their ability to recognize when not to answer. Our zero-shot and abstention experiments reveal significant variation across LLMs, languages, and legal schools of thought. While GPT-4o outperforms all other models in accuracy, Gemini and Fanar demonstrate superior abstention behavior critical for minimizing confident incorrect answers. Notably, all models exhibit a performance drop in Arabic, highlighting the limitations in religious reasoning for languages other than English. To the best of our knowledge, this is the first study to benchmark the efficacy of LLMs for fine-grained Islamic school of thought specific ruling generation and to evaluate abstention for Islamic jurisprudence queries. Our findings underscore the need for task-specific evaluation and cautious deployment of LLMs in religious applications.
- 15 October 2025
- Published by Association for the Advancement of Artificial Intelligence (AAAI) in Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society
- Vol. 8 (1) , 217-226
Open AccessSacred or Synthetic? Evaluating LLM Reliability and Abstention for Religious Questions
Despite the increasing usage of Large Language Models (LLMs) in answering in questions in a variety of domains, their reliability and accuracy remain unexamined for a a plethora of domains, including the religious domains. In this paper, we introduce a novel benchmark FiqhQA focused on the LLM generated Islamic rulings explicitly categorized by the four major Sunni schools of thought, in both Arabic and English. Unlike prior work, which either overlooks the distinctions between religious schools of thought or fails to evaluate abstention behavior, we assess LLMs not only on their accuracy but also on their ability to recognize when not to answer. Our zero-shot and abstention experiments reveal significant variation across LLMs, languages, and legal schools of thought. While GPT-4o outperforms all other models in accuracy, Gemini and Fanar demonstrate superior abstention behavior critical for minimizing confident incorrect answers. Notably, all models exhibit a performance drop in Arabic, highlighting the limitations in religious reasoning for languages other than English. To the best of our knowledge, this is the first study to benchmark the efficacy of LLMs for fine-grained Islamic school of thought specific ruling generation and to evaluate abstention for Islamic jurisprudence queries. Our findings underscore the need for task-specific evaluation and cautious deployment of LLMs in religious applications.
- Open Access
SMAB: MAB based word Sensitivity Estimation Framework and its Applications in Adversarial Text Generation
To understand the complexity of sequence classification tasks, Hahn et al. (2021) proposed sensitivity as the number of disjoint subsets of the input sequence that can each be individually changed to change the output. Though effective, calculating sensitivity at scale using this framework is costly because of exponential time complexity. Therefore, we introduce a Sensitivity-based Multi-Armed Bandit framework (SMAB), which provides a scalable approach for calculating word-level local (sentence-level) and global (aggregated) sensitivities concerning an underlying text classifier for any dataset. We establish the effectiveness of our approach through various applications. We perform a case study on CHECKLIST generated sentiment analysis dataset where we show that our algorithm indeed captures intuitively high and low-sensitive words. Through experiments on multiple tasks and languages, we show that sensitivity can serve as a proxy for accuracy in the absence of gold data. Lastly, we show that guiding perturbation prompts using sensitivity values in adversarial example generation improves attack success rate by 15.58%, whereas using sensitivity as an additional reward in adversarial paraphrase generation gives a 12.00% improvement over SOTA approaches. Warning: Contains potentially offensive content.
Open AccessSMAB: MAB based word Sensitivity Estimation Framework and its Applications in Adversarial Text Generation
To understand the complexity of sequence classification tasks, Hahn et al. (2021) proposed sensitivity as the number of disjoint subsets of the input sequence that can each be individually changed to change the output. Though effective, calculating sensitivity at scale using this framework is costly because of exponential time complexity. Therefore, we introduce a Sensitivity-based Multi-Armed Bandit framework (SMAB), which provides a scalable approach for calculating word-level local (sentence-level) and global (aggregated) sensitivities concerning an underlying text classifier for any dataset. We establish the effectiveness of our approach through various applications. We perform a case study on CHECKLIST generated sentiment analysis dataset where we show that our algorithm indeed captures intuitively high and low-sensitive words. Through experiments on multiple tasks and languages, we show that sensitivity can serve as a proxy for accuracy in the absence of gold data. Lastly, we show that guiding perturbation prompts using sensitivity values in adversarial example generation improves attack success rate by 15.58%, whereas using sensitivity as an additional reward in adversarial paraphrase generation gives a 12.00% improvement over SOTA approaches. Warning: Contains potentially offensive content.
- 15 January 2025
- Published by Springer Nature in Nature Reviews Psychology
- Vol. 4 (2) , 75-80
The promise and pitfalls of generative AI
- 15 January 2025
- Published by Springer Nature in Nature Reviews Psychology
- Vol. 4 (2) , 75-80
The promise and pitfalls of generative AI
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 19114-19137
Women, Infamous, and Exotic Beings: A Comparative Study of Honorific Usages in Wikipedia and LLMs for Bengali and Hindi
Sourabrata Mukherjee, Atharva Mehta, Sougata Saha, Akhil Arora, Monojit Choudhury. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 19114-19137
Open AccessWomen, Infamous, and Exotic Beings: A Comparative Study of Honorific Usages in Wikipedia and LLMs for Bengali and Hindi
Sourabrata Mukherjee, Atharva Mehta, Sougata Saha, Akhil Arora, Monojit Choudhury. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 11047-11065
User Behavior Prediction as a Generic, Robust, Scalable, and Low-Cost Evaluation Strategy for Estimating Generalization in LLMs
Sougata Saha, Monojit Choudhury. Findings of the Association for Computational Linguistics: ACL 2025. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 11047-11065
Open AccessUser Behavior Prediction as a Generic, Robust, Scalable, and Low-Cost Evaluation Strategy for Estimating Generalization in LLMs
Sougata Saha, Monojit Choudhury. Findings of the Association for Computational Linguistics: ACL 2025. 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 8025-8042
Meta-Cultural Competence: Climbing the Right Hill of Cultural Awareness
Sougata Saha, Saurabh Kumar Pandey, Monojit Choudhury. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 8025-8042
Open AccessMeta-Cultural Competence: Climbing the Right Hill of Cultural Awareness
Sougata Saha, Saurabh Kumar Pandey, Monojit Choudhury. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 9158-9176
SMAB: MAB based word Sensitivity Estimation Framework and its Applications in Adversarial Text Generation
Saurabh Kumar Pandey, Sachin Vashistha, Debrup Das, Somak Aditya, Monojit Choudhury. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 9158-9176
Open AccessSMAB: MAB based word Sensitivity Estimation Framework and its Applications in Adversarial Text Generation
Saurabh Kumar Pandey, Sachin Vashistha, Debrup Das, Somak Aditya, Monojit Choudhury. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.
- Open Access
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 4569-4585
Music for All: Representational Bias and Cross-Cultural Adaptability of Music Generation Models
Atharva Mehta, Shivam Chauhan, Amirbek Djanibekov, Atharva Kulkarni, Gus Xia, Monojit Choudhury. Findings of the Association for Computational Linguistics: NAACL 2025. 2025.
- 1 January 2025
- Published by Association for Computational Linguistics (ACL)
- p. 4569-4585
Open AccessMusic for All: Representational Bias and Cross-Cultural Adaptability of Music Generation Models
Atharva Mehta, Shivam Chauhan, Amirbek Djanibekov, Atharva Kulkarni, Gus Xia, Monojit Choudhury. Findings of the Association for Computational Linguistics: NAACL 2025. 2025.