User-Centered Survey on Crafting Effective Prompts for LLMs

A new survey paper on arXiv (2608.07494) examines the principles, taxonomy, and organization of prompts for Large Language Models (LLMs) from a user-centered perspective. The authors argue that while AI tools like ChatGPT and DeepSeek allow users to obtain instant responses by typing requests, crafting clear and well-structured prompts is a systematic and skillful process that can challenge even experienced users. The survey provides actionable guidelines for improving human-LLM interaction, differing from existing surveys that focus primarily on technical principles and application scenarios. The paper is available on arXiv and was announced as a cross-type submission.