Mind the Prompt: Self-adaptive Generation of Task Plan Explanations via LLMs

arXiv cs.AI / 4/25/2026

📰 NewsIdeas & Deep AnalysisModels & Research

共有:

Key Points

The paper argues that while LLMs can produce human-understandable explanations for automated task planning, explanation quality and reliability depend strongly on prompt engineering.
It identifies a gap in systematic understanding of how different stakeholder groups formulate and refine prompts, which limits the ability to automate prompt crafting.
The authors propose COMPASS, a proof-of-concept self-adaptive method that treats prompt engineering as a cognitive and probabilistic decision-making problem.
COMPASS uses a POMDP-based model to infer users’ latent cognitive states (e.g., attention, comprehension, uncertainty) from interaction cues, enabling adaptive generation of explanations and iterative prompt refinements.
Evaluation on two cyber-physical system case studies shows COMPASS can feasibly integrate human cognition and user feedback into automated prompt synthesis for complex task planning.

Abstract

Integrating Large Language Models (LLMs) into complex software systems enables the generation of human-understandable explanations of opaque AI processes, such as automated task planning. However, the quality and reliability of these explanations heavily depend on effective prompt engineering. The lack of a systematic understanding of how diverse stakeholder groups formulate and refine prompts hinders the development of tools that can automate this process. We introduce COMPASS (COgnitive Modelling for Prompt Automated SynthesiS), a proof-of-concept self-adaptive approach that formalises prompt engineering as a cognitive and probabilistic decision-making process. COMPASS models unobservable users' latent cognitive states, such as attention and comprehension, uncertainty, and observable interaction cues as a POMDP, whose synthesised policy enables adaptive generation of explanations and prompt refinements. We evaluate COMPASS using two diverse cyber-physical system case studies to assess the adaptive explanation generation and their qualities, both quantitatively and qualitatively. Our results demonstrate the feasibility of COMPASS integrating human cognition and user profile's feedback into automated prompt synthesis in complex task planning systems.