ChatGPT与AIGC产业链  

ChatGPT and AIGC Industry Chain

作　　者：史占中[1] 郑世民 蒋越 SHI Zhanzhong;ZHENG Shimin;JIANG Yue(Antai College of Economics&Management,Shanghai Jiao Tong University,Shanghai 200030,China)

机构地区：[1]上海交通大学安泰经济与管理学院,上海200030

出　　处：《上海管理科学》2023年第2期12-14,共3页Shanghai Management Science

摘　　要：ChatGPT是一种生成式预训练模型,结合了人工反馈的强化学习技术,运用大量计算资源和海量数据,实现了对人类语言的理解和学习,存储了海量的知识。对语言的深入理解和广泛的知识存储,以及强大的内容生成能力,将会给搜索引擎等领域带来巨大的冲击,并且为AIGC特别是商业化的发展带来了新的机会。尽管ChatGPT有相对较高的技术壁垒,但依托国内人工智能领域完备的基础设施和充分的人才储备,我们认为在AIGC产业链方面仍然可以把握更多的发展机会。ChatGPT is a pre-trained chat model using a simple network architecture,the Transformer,combined with reinforcement learning from human feedback,using large computational resources and massive data to achieve understanding and learning of human language and store massive knowledge.The deep understanding and extensive knowledge of language,as well as the powerful content generation capability,will bring a big impact to fields such as search engines,and bring new opportunities for the development,especially commercialization,of AIGC.Although ChatGPT has considerable technical barriers,China has a complete infrastructure and sufficient talent pool in AI,so domestic academia and industry are still full of opportunities in AIGC.We show that the ChatGPT generalizes well to other tasks by applying it successfully to industries both with large and limited training data.

关 键 词：ChatGPT 自然语言处理 强化学习 AIGC 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]