Publications (17)
Probabilities of Chat LLMs Are Miscalibrated but Still Predict Correctness on Multiple-Choice Q&A
Benjamin Plaut, Nguyen X. Khanh, Tu Trinh
We study 15 large language models (LLMs) fine-tuned for chat and find that their maximum softmax probabilities (MSPs) are consistently miscalibrated on multiple-choice Q&A. However…
Communication Complexity of Discrete Fair Division
Benjamin Plaut, Tim Roughgarden
We initiate the study of the communication complexity of fair division with indivisible goods. We focus on some of the most well-studied fairness notions (envy-freeness, proportion…
Safety Training Persists Through Helpfulness Optimization in LLM Agents
Benjamin Plaut
Safety post-training has been studied extensively in single-step "chat" settings where safety typically refers to refusing harmful requests. We study an "agentic" (i.e., multi-step…
Hardness of the Pricing Problem for Chains in Barter Exchanges
Benjamin Plaut, John P. Dickerson, Tuomas Sandholm
Kidney exchange is a barter market where patients trade willing but medically incompatible donors. These trades occur via cycles, where each patient-donor pair both gives and recei…
Markets Beyond Nash Welfare for Leontief Utilities
Ashish Goel, Reyna Hulett, Benjamin Plaut
We study the allocation of divisible goods to competing agents via a market mechanism, focusing on agents with Leontief utilities. The majority of the economics and mechanism desig…
Check Yourself Before You Wreck Yourself: Selectively Quitting Improves LLM Agent Safety
Vamshi Krishna Bonagiri, Ponnurangam Kumaragurum, Khanh Nguyen +1
As Large Language Model (LLM) agents increasingly operate in complex environments with real-world consequences, their safety becomes critical. While uncertainty quantification is w…
Counteracting Inequality in Markets via Convex Pricing
Ashish Goel, Benjamin Plaut
We study market mechanisms for allocating divisible goods to competing agents with quasilinear utilities. For \emph{linear} pricing (i.e., the cost of a good is proportional to the…
Learning When Not to Learn: Risk-Sensitive Abstention in Bandits with Unbounded Rewards
Sarah Liaw, Benjamin Plaut
In high-stakes AI applications, even a single action can cause irreparable damage. However, nearly all of sequential decision-making theory assumes that all errors are recoverable…
Almost Envy-Freeness with General Valuations
Benjamin Plaut, Tim Roughgarden
The goal of fair division is to distribute resources among competing players in a "fair" way. Envy-freeness is the most extensively studied fairness notion in fair division. Envy-f…
Safe Learning Under Irreversible Dynamics via Asking for Help
Benjamin Plaut, Juan Liévano-Karim, Hanlin Zhu +1
Most learning algorithms with formal regret guarantees essentially rely on trying all possible behaviors, which is problematic when some errors cannot be recovered from. Instead, w…
Markets for Public Decision-making
Nikhil Garg, Ashish Goel, Benjamin Plaut
A public decision-making problem consists of a set of issues, each with multiple possible alternatives, and a set of competing agents, each with a preferred alternative for each is…
Position-Indexed Formulations for Kidney Exchange
John P. Dickerson, David F. Manlove, Benjamin Plaut +2
A kidney exchange is an organized barter market where patients in need of a kidney swap willing but incompatible donors. Determining an optimal set of exchanges is theoretically an…
Getting By Goal Misgeneralization With a Little Help From a Mentor
Tu Trinh, Mohamad H. Danesh, Nguyen X. Khanh +1
While reinforcement learning (RL) agents often perform well during training, they can struggle with distribution shift in real-world deployments. One particularly severe risk of di…
Optimal Nash Equilibria for Bandwidth Allocation
Benjamin Plaut
In bandwidth allocation, competing agents wish to transmit data along paths of links in a network, and each agent's utility is equal to the minimum bandwidth she receives among all…
Avoiding Catastrophe in Online Learning by Asking for Help
Benjamin Plaut, Hanlin Zhu, Stuart Russell
Most learning algorithms with formal regret guarantees assume that all mistakes are recoverable and essentially rely on trying all possible behaviors. This approach is problematic…
YRC-Bench: A Benchmark for Learning to Coordinate with Experts
Mohamad H. Danesh, Nguyen X. Khanh, Tu Trinh +1
When deployed in the real world, AI agents will inevitably face challenges that exceed their individual capabilities. A critical component of AI safety is an agent's ability to rec…
Almost Envy-free Repeated Matching in Two-sided Markets
Sreenivas Gollapudi, Kostas Kollias, Benjamin Plaut
A two-sided market consists of two sets of agents, each of whom have preferences over the other (Airbnb, Upwork, Lyft, Uber, etc.). We propose and analyze a repeated matching probl…