ValuePilot: A Two-Phase Framework for Value-Driven Decision-Making

2025-03-06

AI Summary

This work addresses the challenge of enabling large language models to make personalized, value-aligned decisions on unseen tasks. We propose a value-driven two-stage framework: first, a Data Generation Tool (DGT) synthesizes high-fidelity, filterable scenario datasets guided by multidimensional value specifications and validated via human-AI collaborative verification; second, a Decision Modeling Module (DMM) is trained to explicitly identify latent values, assess action feasibility, and model and resolve multidimensional value conflicts through structured tradeoff reasoning. To our knowledge, this is the first paradigm to deeply integrate multidimensional value modeling into the core decision-making process, enabling fine-grained alignment with human preferences. Experiments under human value constraints demonstrate that DMM achieves significantly higher decision consistency than Claude-3.5-Sonnet, Gemini-2-Flash, Llama-3.1-405B, and GPT-4o, validating both its effectiveness and strong cross-task generalization capability.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Machine Learning: Large Multimodal Models (LMMs)
• Philosophy and Ethics of AI: Morality & Value-based AI

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• 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

Abstract

Despite recent advances in artificial intelligence (AI), it poses challenges to ensure personalized decision-making in tasks that are not considered in training datasets. To address this issue, we propose ValuePilot, a two-phase value-driven decision-making framework comprising a dataset generation toolkit DGT and a decision-making module DMM trained on the generated data. DGT is capable of generating scenarios based on value dimensions and closely mirroring real-world tasks, with automated filtering techniques and human curation to ensure the validity of the dataset. In the generated dataset, DMM learns to recognize the inherent values of scenarios, computes action feasibility and navigates the trade-offs between multiple value dimensions to make personalized decisions. Extensive experiments demonstrate that, given human value preferences, our DMM most closely aligns with human decisions, outperforming Claude-3.5-Sonnet, Gemini-2-flash, Llama-3.1-405b and GPT-4o. This research is a preliminary exploration of value-driven decision-making. We hope it will stimulate interest in value-driven decision-making and personalized decision-making within the community.

Problem

• Addresses personalized decision-making in AI tasks not covered by training datasets.
• Proposes ValuePilot, a framework for value-driven decision-making with two modules.
• Ensures alignment with human decisions by learning and navigating value dimensions.

Innovation

• Two-phase framework for value-driven decision-making
• Dataset generation toolkit with automated filtering
• Decision-making module learning scenario values