大模型驱动地震科普数字人的实践与探索

Exploration and Use of Large Model-driven Digital Humans for Popularizing Earthquake Science

我国地震多发,地震灾害严重,积极推动防震减灾科普对增强公众防震和主动减灾意识具有重要意义。目前,地震科普内容、表现形式、丰富性和趣味性等方面仍存在很大不足,科普成效需要增强。利用人工智能技术和语言大模型,开发地震科普大模型,创新地震科普新方式,可增强科普互动性和趣味性,丰富科普内容。地震科学国际数据中心基于阿里“通义千问”开源模型打造了地震垂直领域大模型QuakeGPT,该模型学习大量地震相关专业知识,可通过问答服务,提供更丰富的地震知识,回答更准确。QuakeGPT模型通过微信“小Q”提供地震知识和防灾建议,增强了公众的参与度和学习效果,显著提升公众对地震科学的认识和应对能力。

Earthquakes in China frequently cause severe damage,underscoring the importance of actively promoting science education to enhance public awareness on earthquake prevention and disaster reduction and proactive disaster mitigation.However,significant gaps in the content,presentation,and richness of current earthquake science-related education exist as well as a lack of public engagement in such efforts,necessitating improvement in all aspects.By leveraging artificial intelligence and large language models,we can develop advanced educational tools to create new innovative methods for enhancing public interactivity and engagement and content richness in earthquake science-related education.The International Seismological Data Center has created QuakeGPT,a large model for the vertical earthquake domain based on Alibaba’s opensource model“Tongyi Qianwen.”This model has acquired extensive professional knowledge related to earthquakes and provides richer,more accurate information through question-and-answer services.QuakeGPT offers earthquake knowledge and disaster prevention advice via the WeChat“XiaoQ”platform,increasing public engagement and learning effectiveness and thereby significantly enhancing the general public’s knowledge on earthquake science and their ability to respond to earthquakes.