LLM Agents Are Hypersensitive to Nudges

TL;DR

This paper investigates how large language models (LLMs) are highly sensitive to nudges in decision-making scenarios, revealing their susceptibility, divergence from human behavior, and the impact of prompting strategies, emphasizing the need for behavioral testing before deployment.

Contribution

The study provides a comprehensive analysis of LLM decision-making under nudges, highlighting their hypersensitivity and exploring prompt-based mitigation strategies, which is a novel examination of LLM behavioral robustness.

Findings

LLMs are more susceptible to nudges than humans.

Prompt strategies can shift LLM decision distributions.

Nudges can significantly affect LLM performance and choices.

Abstract

LLMs are being set loose in complex, real-world environments involving sequential decision-making and tool use. Often, this involves making choices on behalf of human users. However, not much is known about the distribution of such choices, and how susceptible they are to different choice architectures. We perform a case study with a few such LLM models on a multi-attribute tabular decision-making problem, under canonical nudges such as the default option, suggestions, and information highlighting, as well as additional prompting strategies. We show that, despite superficial similarities to human choice distributions, such models differ in subtle but important ways. First, they show much higher susceptibility to the nudges. Second, they diverge in points earned, being affected by factors like the idiosyncrasy of available prizes. Third, they diverge in information acquisition strategies: e.g. incurring substantial cost to reveal too much information, or selecting without revealing any. Moreover, we show that simple prompt strategies like zero-shot chain of thought (CoT) can shift the choice distribution, and few-shot prompting with human data can induce greater alignment. Yet, none of these methods resolve the sensitivity of these models to nudges. Finally, we show how optimal nudges optimized with a human resource-rational model can similarly increase LLM performance for some models. All these findings suggest that behavioral tests are needed before deploying models as agents or assistants acting on behalf of users in complex environments.