1

From Delegates to Trustees: How Optimizing for Long-Term Interests Shapes Bias and Alignment in LLMs

Examining how optimizing LLMs for long-term interests shapes bias and alignment.

Suyash Fulay, Jocelyn Zhu, Michiel Bakker

Agora: Teaching the Skill of Consensus-Finding with AI Personas Grounded in Human Voice

Teaching consensus-finding with AI personas grounded in human voice.

Suyash Fulay, Prerna Ravi, Om Gokhale, Eugene Yi, Michiel Bakker, Deb Roy

Using LLMs to Raise Missing Perspectives in Policy Deliberations

Using LLMs to surface missing perspectives in policy deliberations. Oral talk (top 10%).

Suyash Fulay, Dimitra Dimitrakopoulou, Deb Roy

AI and Collective Decisions: Strengthening Legitimacy and Losers' Consent

Exploring how AI can strengthen legitimacy and losers’ consent in collective decision-making processes.

Suyash Fulay, Prerna Ravi, Emily Kubin, Shreshtha Mohanty, Michiel Bakker, Deb Roy

Coalesce: An Accessible Mixed-Initiative System for Designing Community-Centric Questionnaires

A mixed-initiative system for designing community-centric questionnaires.

Cassandra Overney, Daniel Kessler, Suyash Fulay, Mahmood Jasin, Deb Roy

On the Relationship between Truth and Political Bias in Language Models

Investigating fundamental trade-offs between truthfulness and political bias in language models.

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

ConGraT: Self-Supervised Contrastive Pretraining for Joint Graph and Text Embeddings

We propose a contrastive method to jointly learn graph and text embeddings.

William Brannon, Wonjune Kang, Suyash Fulay, Hang Jiang, Brandon Roy, Jad Kabbara, Deb Roy

ConGraT: Self-Supervised Contrastive Pretraining for Joint Graph and Text Embeddings

Divergences in Following Patterns between Influential Twitter Users and Their Audiences

Proposing graph-based metrics to understand people’s preferences on social networks.

Suyash Fulay, Nabeel Gillani, Deb Roy