AI RESEARCH

UtilityMax Prompting: A Formal Framework for Multi-Objective Large Language Model Optimization

arXiv CS.AI

ArXi:2603.11583v1 Announce Type: cross The success of a Large Language Model (LLM) task depends heavily on its prompt. Most use-cases specify prompts using natural language, which is inherently ambiguous when multiple objectives must be simultaneously satisfied. In this paper we