publications

2026

1. ArXiv 2026

   Position: A Roadmap to Impactful Pluralistic Alignment Research

   Elinor Poole-Dayan, Jillian Fisher, Atoosa Kasirzadeh, Jacob Andreas, Mitchell Gordon, and Michiel Bakker

   Jul 2026

   Abs arXiv PDF Website

   Pluralistic value alignment—the goal of building AI systems that represent and serve diverse human values and perspectives—has emerged as an active research agenda. Yet, there’s no public evidence that it has shaped the training or evaluation of the AI systems people actually use. While some frontier labs describe some form of pluralistic behavior from their models in their constitutions and model specs, none name pluralism as a goal, and as of this writing, their public model cards, system cards, and evaluations don’t clearly indicate that production models are explicitly trained or tested for it. This goes against the primary motivations and goals of pluralistic alignment, which revolve around making a positive difference in the models serving billions of users worldwide. If our efforts never reach beyond the research sphere and into deployed systems, the field will ultimately fail to achieve these goals. We argue that the pluralistic alignment research community, the researchers advancing this agenda across academia, non-profits, and industry labs, should focus on supporting impact and adoption in deployed, widely-used AI systems. In this paper, we provide evidence for the adoption problem, present three main reasons behind it, and discuss three corresponding areas for future research to address it: (i) The primary justifications for pluralistic alignment so far have been normative or speculative, with few studies showing empirically how pluralistic AI benefits users or society concretely. We need empirical foundations for pluralistic AI. (ii) The pluralistic alignment research community has not settled when pluralistic behavior is warranted or what an ideal pluralistic response looks like, so there is no concrete goal for developers of widely used models to operationalize. We need an account of when pluralism is warranted and how it should look in practice, concrete enough for a model spec or constitution to state as policy. (iii) Current methods, such as multi-model collaboration and pluralistic reinforcement learning, trade off against other desiderata of LLMs in ways that are largely unmeasured, and existing metrics are not “hill-climbable,” so there is nothing a model developer could adopt and optimize today. We need trade-off-aware evaluations and methods that meet the requirements of production systems. This position paper serves as a collective call to action for the pluralistic alignment research community: progress requires moving beyond normative justification toward empirical foundations, a concrete account of ideal pluralistic behavior, and practical methodologies and evaluations built for adoption.

2. ICLR 2026

   Benchmarking Overton Pluralism in LLMs

   Elinor Poole-Dayan, Jiayi Wu, Taylor Sorensen, Jiaxin Pei, and Michiel A. Bakker

   In The Fourteenth International Conference on Learning Representations, Apr 2026

   Abs arXiv Code Website Dataset

   We introduce a novel framework for measuring Overton pluralism in LLMs–the extent to which diverse viewpoints are represented in model outputs. We (i) formalize Overton pluralism as a set coverage metric (OvertonScore), (ii) conduct a large-scale U.S.-representative human study (N = 1209; 60 questions; 8 LLMs), and (iii) develop an automated benchmark that closely reproduces human judgments. On average, models achieve OvertonScores of 0.35–0.41, with DeepSeek V3 performing best; yet all models remain far below the theoretical maximum of 1.0, revealing substantial headroom for improvement. Because repeated large-scale human studies are costly and slow, scalable evaluation tools are essential for model development. Hence, we propose an automated benchmark that achieves high rank correlation with human judgments (ρ=0.88), providing a practical proxy without replacing human assessment. By turning pluralistic alignment from a normative aim into a measurable benchmark, our work establishes a foundation for systematic progress toward more pluralistic LLMs.

3. AAAI 2026

   LLM Targeted Underperformance Disproportionately Impacts Vulnerable Users

   Elinor Poole-Dayan, Deb Roy, and Jad Kabbara

   In Proceedings of the 40th AAAI Conference on Artificial Intelligence (AAAI-26), Jan 2026

   Covered by MIT News.

   Abs DOI arXiv PDF Poster

   While state-of-the-art large language models (LLMs) have shown impressive performance on many tasks, systematically evaluating undesirable behaviors of these models remains critical. In this work, we investigate how the quality of LLM responses changes in terms of information accuracy, truthfulness, and refusals depending on three user traits: English proficiency, education level, and country of origin. We present extensive experimentation on three state-of-the-art LLMs and two different datasets targeting truthfulness and factuality. Our findings suggest that undesirable behaviors in state-of-the-art LLMs occur disproportionately more for users with lower English proficiency, of lower education status, and originating from outside the US, rendering these models unreliable sources of information towards their most vulnerable users.

2025

1. EMNLP 2025

   Computational Analysis of Conversation Dynamics through Participant Responsivity

   Margaret Hughes , Brandon Roy, Elinor Poole-Dayan, Deb Roy, and Jad Kabbara

   In Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing, Nov 2025

   Abs arXiv PDF Demo

   Growing literature explores toxicity and polarization in discourse, with comparatively less work on characterizing what makes dialogue prosocial and constructive. We explore conversational discourse and investigate a method for characterizing its quality built upon the notion of “responsivity”—whether one person’s conversational turn is responding to a preceding turn. We develop and evaluate methods for quantifying responsivity—first through semantic similarity of speaker turns, and second by leveraging state-of-the-art large language models (LLMs) to identify the relation between two speaker turns. We evaluate both methods against a ground truth set of human-annotated conversations. Furthermore, selecting the better performing LLM-based approach, we characterize the nature of the response—whether it responded to that preceding turn in a substantive way or not. We view these responsivity links as a fundamental aspect of dialogue but note that conversations can exhibit significantly different responsivity structures. Accordingly, we then develop conversation-level derived metrics to address various aspects of conversational discourse. We use these derived metrics to explore other conversations and show that they support meaningful characterizations and differentiations across a diverse collection of conversations.

2. COLM 2025 NLP4Dem

   An AI-Powered Framework for Analyzing Collective Idea Evolution in Deliberative Assemblies

   Elinor Poole-Dayan, Deb Roy, and Jad Kabbara

   Accepted as an Oral Presentation at the COLM 2025 NLP4Democracy Workshop , Oct 2025

   Abs arXiv Slides

   In an era of increasing societal fragmentation, political polarization, and erosion of public trust in institutions, representative deliberative assemblies are emerging as a promising democratic forum for developing effective policy outcomes on complex global issues. Despite theoretical attention, there remains limited empirical work that systematically traces how specific ideas evolve, are prioritized, or are discarded during deliberation to form policy recommendations. Addressing these gaps, this work poses two central questions: (1) How might we trace the evolution and distillation of ideas into concrete recommendations within deliberative assemblies? (2) How does the deliberative process shape delegate perspectives and influence voting dynamics over the course of the assembly? To address these questions, we develop LLM-based methodologies for empirically analyzing transcripts from a tech-enhanced in-person deliberative assembly. The framework identifies and visualizes the space of expressed suggestions. We also empirically reconstruct each delegate’s evolving perspective throughout the assembly. Our methods contribute novel empirical insights into deliberative processes and demonstrate how LLMs can surface high-resolution dynamics otherwise invisible in traditional assembly outputs.

3. MIT 2025

   From Dialogue to Decision: An LLM-Powered Framework for Analyzing Collective Idea Evolution and Voting Dynamics in Deliberative Assemblies

   Elinor Poole-Dayan

   May 2025

   Advisor: Deb Roy

   PDF
4. NAACL 2025 WNU

   Applying Large Language Models to Characterize Public Narratives

   Elinor Poole-Dayan^*, Daniel T Kessler^*, Hannah Chiou, Margaret Hughes, Emily S. Lin, Marshall Ganz, and Deb Roy

   Accepted at the NAACL 2025 Workshop on Narrative Understanding (WNU) , May 2025

   Abs arXiv Poster Website Demo

   Public Narratives (PNs) are key tools for leadership development and civic mobilization, yet their systematic analysis remains challenging due to their subjective interpretation and the high cost of expert annotation. In this work, we propose a novel computational framework that leverages large language models (LLMs) to automate the qualitative annotation of public narratives. Using a codebook we co-developed with subject-matter experts, we evaluate LLM performance against that of expert annotators. Our work reveals that LLMs can achieve near-human-expert performance, achieving an average F1 score of 0.80 across 8 narratives and 14 codes. We then extend our analysis to empirically explore how PN framework elements manifest across a larger dataset of 22 stories. Lastly, we extrapolate our analysis to a set of political speeches, establishing a novel lens in which to analyze political rhetoric in civic spaces. This study demonstrates the potential of LLM-assisted annotation for scalable narrative analysis and highlights key limitations and directions for future research in computational civic storytelling.

2024

1. EMNLP 2024

   On the Relationship between Truth and Political Bias in Language Models

   Suyash Fulay, William Brannon, Shrestha Mohanty, Cassandra Overney, Elinor Poole-Dayan, Deb Roy, and Jad Kabbara

   In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, Nov 2024

   Covered by MIT News, Communications of the ACM (AI + ML).

   Abs arXiv PDF Code

   Language model alignment research often attempts to ensure that models are not only helpful and harmless, but also truthful and unbiased. However, optimizing these objectives simultaneously can obscure how improving one aspect might impact the others. In this work, we focus on analyzing the relationship between two concepts essential in both language model alignment and political science: truthfulness and political bias. We train reward models on various popular truthfulness datasets and subsequently evaluate their political bias. Our findings reveal that optimizing reward models for truthfulness on these datasets tends to result in a left-leaning political bias. We also find that existing open-source reward models (i.e., those trained on standard human preference datasets) already show a similar bias and that the bias is larger for larger models. These results raise important questions about the datasets used to represent truthfulness, potential limitations of aligning models to be both truthful and politically unbiased, and what language models capture about the relationship between truth and politics.

2023

1. NeurIPS 2023

   Are Diffusion Models Vision-And-Language Reasoners?

   Benno Krojer, Elinor Poole-Dayan, Vikram Voleti, Chris Pal, and Siva Reddy

   In Advances in Neural Information Processing Systems, Nov 2023

   Abs arXiv PDF Code

   Text-conditioned image generation models have recently shown immense qualitative success using denoising diffusion processes. However, unlike discriminative vision-and-language models, it is a non-trivial task to subject these diffusion-based generative models to automatic fine-grained quantitative evaluation of high-level phenomena such as compositionality.Towards this goal, we perform two innovations. First, we transform diffusion-based models (in our case, Stable Diffusion) for any image-text matching (ITM) task using a novel method called DiffusionITM.Second, we introduce the Generative-Discriminative Evaluation Benchmark (GDBench) benchmark with 7 complex vision-and-language tasks, bias evaluation and detailed analysis.We find that Stable Diffusion + DiffusionITM is competitive on many tasks and outperforms CLIP on compositional tasks like like CLEVR and Winoground.We further boost its compositional performance with a transfer setup by fine-tuning on MS-COCO while retaining generative capabilities. We also measure the stereotypical bias in diffusion models, and find that Stable Diffusion 2.1 is, for the most part, less biased than Stable Diffusion 1.5.Overall, our results point in an exciting direction bringing discriminative and generative model evaluation closer. We will release code and benchmark setup soon.

2022

1. ACL 2022

   An Empirical Survey of the Effectiveness of Debiasing Techniques for Pre-trained Language Models

   Nicholas Meade, Elinor Poole-Dayan, and Siva Reddy

   In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), May 2022

   Abs DOI arXiv PDF Code Website

   Recent work has shown pre-trained language models capture social biases from the large amounts of text they are trained on. This has attracted attention to developing techniques that mitigate such biases. In this work, we perform an empirical survey of five recently proposed bias mitigation techniques: Counterfactual Data Augmentation (CDA), Dropout, Iterative Nullspace Projection, Self-Debias, and SentenceDebias. We quantify the effectiveness of each technique using three intrinsic bias benchmarks while also measuring the impact of these techniques on a model’s language modeling ability, as well as its performance on downstream NLU tasks. We experimentally find that: (1) Self-Debias is the strongest debiasing technique, obtaining improved scores on all bias benchmarks; (2) Current debiasing techniques perform less consistently when mitigating non-gender biases; And (3) improvements on bias benchmarks such as StereoSet and CrowS-Pairs by using debiasing strategies are often accompanied by a decrease in language modeling ability, making it difficult to determine whether the bias mitigation was effective.