【作 者】：
王 珊，肖义墙，潘亦宁，彭 唯
【关 键 词 】：
生成式人工智能； 家庭教育能力； 积极家庭心理； 理论模型； 系统实践
【栏 目】：
教育数字化
【中文摘要】：
当下中国家庭面临着教育知识经验化与碎片化、密集教养方式与过度教育焦虑恶性循环以及个性化家庭教育指导服务匮乏的三重困境。为破解家庭教育的实践困境、提升现代家庭教育能力、促进积极家庭心理成长，研究构建了生成式人工智能助力家庭教育的理论模型，并在GLM-4大模型的基础上运用提示词工程、检索增强生成以及有监督微调技术开发了家庭教育生成式人工智能系统。从实践效果来看，家庭教育生成式人工智能系统具备卓越的情境适应性和对特定任务极强的泛化性能。家庭教育生成式人工智能系统能精准分析和解决各种复杂的家庭教育问题，为家长提供科学系统的育儿知识和正面管教指导，缓解家长教育焦虑，并为不同社会经济背景的家庭提供个性化的教育指导服务，有利于促进我国家庭教育的现代化发展。
【英文摘要】：
Nowadays, Chinese families face the triple dilemma of empirical and fragmented educational knowledge, the vicious circle of intensive parenting style and excessive educational anxiety, and the lack of personalized family education guidance services. In order to solve the practical dilemma of family education, improve the ability of modern family education, and promote the positive psychological growth of families, the study constructed a theoretical model of generative artificial intelligence to assist family education, and developed a generative artificial intelligence system for family education on the basis of the GLM-4 model by using prompt engineering, retrieval augmented generation, and supervised fine-tuning. From the practical results, the generative artificial intelligence system for family education has excellent contextual adaptability and strong generalization performance for specific tasks. The system for family education can accurately analyse and solve various complex family education problems, provide parents with scientific and systematic parenting knowledge and positive discipline guidance, alleviate parents' educational anxiety, and provide personalized educational guidance services for families with different socioeconomic backgrounds, which is conducive to promoting the modernization of family education in China.