Cash or Comfort? How LLMs Value Your Inconvenience

2025-06-20

AI Summary

This study investigates the reliability of large language models (LLMs) as personal utility proxies under conflicts between financial incentives and subjective discomfort—including physical states (e.g., walking, hunger, pain) and psychological states (e.g., waiting). Using a behavioral economics framework, we design standardized discomfort–compensation trade-off tasks and conduct multi-round prompt engineering with cross-model comparative experiments. For the first time, we systematically quantify monetary valuations of diverse discomfort types across 12 mainstream LLMs. Four critical deficiencies emerge: (1) extreme inter-model variance in valuations; (2) high sensitivity of decisions to minor prompt perturbations; (3) widespread acceptance of trivial compensation for substantial discomfort; and (4) unwarranted rejection of zero-cost beneficial actions. These findings demonstrate that current LLMs lack robust, interpretable utility modeling capabilities—rendering them unsuitable for deployment in sensitive, high-stakes personal decision-making contexts.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Humans and AI: Learning Human Values and Preferences
• Reasoning under Uncertainty: Decision/Utility Theory

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large Language Models (LLMs) are increasingly proposed as near-autonomous artificial intelligence (AI) agents capable of making everyday decisions on behalf of humans. Although LLMs perform well on many technical tasks, their behaviour in personal decision-making remains less understood. Previous studies have assessed their rationality and moral alignment with human decisions. However, the behaviour of AI assistants in scenarios where financial rewards are at odds with user comfort has not yet been thoroughly explored. In this paper, we tackle this problem by quantifying the prices assigned by multiple LLMs to a series of user discomforts: additional walking, waiting, hunger and pain. We uncover several key concerns that strongly question the prospect of using current LLMs as decision-making assistants: (1) a large variance in responses between LLMs, (2) within a single LLM, responses show fragility to minor variations in prompt phrasing (e.g., reformulating the question in the first person can considerably alter the decision), (3) LLMs can accept unreasonably low rewards for major inconveniences (e.g., 1 Euro to wait 10 hours), and (4) LLMs can reject monetary gains where no discomfort is imposed (e.g., 1,000 Euro to wait 0 minutes). These findings emphasize the need for scrutiny of how LLMs value human inconvenience, particularly as we move toward applications where such cash-versus-comfort trade-offs are made on users' behalf.

Problem

• Assessing LLMs' decision-making in cash-versus-comfort trade-offs
• Quantifying LLMs' pricing of user discomforts like pain and waiting
• Examining LLMs' inconsistency and irrationality in valuing inconvenience

Innovation

• Quantify LLM-assigned prices for user discomforts
• Assess variance and fragility in LLM responses
• Evaluate LLM reward acceptance for inconveniences