Colloquium - Bhuwan Dhinga, "Language Models as Epistemic Interfaces"

Abstract: 
Chatbots are rapidly becoming the primary interface through which people consume information. By replacing search and synthesis with direct answers, these systems reduce cognitive effort and enable powerful new forms of knowledge interaction. At the same time, they obscure important epistemic signals: where the information comes from, how uncertain it is, and how it was produced. Paradoxically, as systems become more capable, users are less likely to question them – amplifying these concerns.

In this talk, I will present recent work from my lab on designing mechanisms that support more reliable use of language models for information consumption. First, I will introduce a unified framework for calibrating long-form generation, where correctness and confidence are treated as distributions rather than binary outcomes. Next, I will present a retrieval-free approach to knowledge attribution that enables models to reliably cite documents encountered during continual pretraining. Finally, I will discuss how coding agents can externalize long-context processing into explicit, executable interactions with file systems and tools, providing an alternative to purely attention-based reasoning. Together, these works explore how to shift language models from opaque answer synthesizers toward more reliable interfaces to knowledge.

Bio: 
Bhuwan Dhingra is an Assistant Professor of Computer Science at Duke University and a Research Scientist at Apple. He has also spent time at Google DeepMind as part of the post-training team for the Gemini foundation models. His research focuses on improving the trustworthiness and efficiency of large language models for knowledge-intensive tasks. He has served as a Senior Area Chair for ACL and an Area Chair for NeurIPS, ICLR, ICML, and EMNLP. He received his bachelor’s degree from IIT Kanpur and his Ph.D. from Carnegie Mellon University. His research is supported by grants from the NSF, Amazon, Procter & Gamble, and the Learning Engineering Virtual Institute. He received the Amazon Research Award in 2021.