Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Luo, Yitong, Chen, Ziang, Lam, Hou Hei, zhan, Jiayu, Wang, Junqi, Zhang, Zhenliang, Feng, Xue
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2512.13716
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909964573868032
author Luo, Yitong
Chen, Ziang
Lam, Hou Hei
zhan, Jiayu
Wang, Junqi
Zhang, Zhenliang
Feng, Xue
author_facet Luo, Yitong
Chen, Ziang
Lam, Hou Hei
zhan, Jiayu
Wang, Junqi
Zhang, Zhenliang
Feng, Xue
contents Personalized decision-making is essential for human-AI interaction, enabling AI agents to act in alignment with individual users' value preferences. As AI systems expand into real-world applications, adapting to personalized values beyond task completion or collective alignment has become a critical challenge. We address this by proposing a value-driven approach to personalized decision-making. Human values serve as stable, transferable signals that support consistent and generalizable behavior across contexts. Compared to task-oriented paradigms driven by external rewards and incentives, value-driven decision-making enhances interpretability and enables agents to act appropriately even in novel scenarios. We introduce ValuePilot, a two-phase framework consisting of a dataset generation toolkit (DGT) and a decision-making module (DMM). DGT constructs diverse, value-annotated scenarios from a human-LLM collaborative pipeline. DMM learns to evaluate actions based on personal value preferences, enabling context-sensitive, individualized decisions. When evaluated on previously unseen scenarios, DMM outperforms strong LLM baselines, including GPT-5, Claude-Sonnet-4, Gemini-2-flash, and Llama-3.1-70b, in aligning with human action choices. Our results demonstrate that value-driven decision-making is an effective and extensible engineering pathway toward building interpretable, personalized AI agents.
format Preprint
id arxiv_https___arxiv_org_abs_2512_13716
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ValuePilot: A Two-Phase Framework for Value-Driven Decision-Making
Luo, Yitong
Chen, Ziang
Lam, Hou Hei
zhan, Jiayu
Wang, Junqi
Zhang, Zhenliang
Feng, Xue
Artificial Intelligence
Personalized decision-making is essential for human-AI interaction, enabling AI agents to act in alignment with individual users' value preferences. As AI systems expand into real-world applications, adapting to personalized values beyond task completion or collective alignment has become a critical challenge. We address this by proposing a value-driven approach to personalized decision-making. Human values serve as stable, transferable signals that support consistent and generalizable behavior across contexts. Compared to task-oriented paradigms driven by external rewards and incentives, value-driven decision-making enhances interpretability and enables agents to act appropriately even in novel scenarios. We introduce ValuePilot, a two-phase framework consisting of a dataset generation toolkit (DGT) and a decision-making module (DMM). DGT constructs diverse, value-annotated scenarios from a human-LLM collaborative pipeline. DMM learns to evaluate actions based on personal value preferences, enabling context-sensitive, individualized decisions. When evaluated on previously unseen scenarios, DMM outperforms strong LLM baselines, including GPT-5, Claude-Sonnet-4, Gemini-2-flash, and Llama-3.1-70b, in aligning with human action choices. Our results demonstrate that value-driven decision-making is an effective and extensible engineering pathway toward building interpretable, personalized AI agents.
title ValuePilot: A Two-Phase Framework for Value-Driven Decision-Making
topic Artificial Intelligence
url https://arxiv.org/abs/2512.13716