摘要
透明度义务是可信人工智能的基本义务,是技术性正当程序的构成要素,是实现可视正义目标的关键性责任机制。作为弱人工智能向强人工智能过渡的ChatGPT具有生成结构化内容的能力,它在提高学术能力的同时也会遮蔽学术失信行为,对学术诚信构成前所未有的威胁。为确保人工智能时代的学术诚信,必须设立透明度义务,适用学术诚信规范与可信人工智能准则,共同规制ChatGPT介入学术论文创作活动,以保障学术诚信、保证学术公平和学术责任。学术界对ChatGPT能否介入学术论文创作所持的开放使用观、合理使用观和禁止使用观,以及关于ChatGPT生成内容可版权性的主体说、客体说和二元创作主体说,都直接影响到透明度义务的履行标准、履行方式和责任形式。关于透明度义务履行的密度标准有“鱼缸透明度”模式与“合理透明度”模式。无论选择哪一种观念、学说和模式履行透明度义务,在内容上取决于学术论文的目的和用途,制度上取决于学术规范,程序上是为了可视正义。
The transparency obligation is a fundamental obligation of trustworthy artificial intelligence,a constituent element of technological due process,and a key responsibility mechanism for achieving the goal of visual justice.As a transition from artificial narrow intelligence to artificial general intelligence,ChatGPT has the ability to generate structured content.While improving academic capabilities,it can also mask academic dishonesty,posing an unprecedented threat to academic integrity.In order to ensure academic integrity in the era of artificial intelligence,it is necessary to establish transparency obligation,apply academic integrity norms and trustworthy AI guidelines,and jointly regulate ChatGPT’s intervention in academic paper creation activities to ensure academic integrity,academic fairness,and academic responsibility.The academic community’s views on whether ChatGPT can be used in the creation of academic papers,such as open use,fair use,and prohibited use,as well as the subject,object,and dual creative subject theories on the copyright of ChatGPT generated content,have a direct impact on the fulfillment standards,fulfillment methods,and forms of responsibility for transparency obligation.The density standards for fulfilling transparency obligation include the“fishbowl transparency”mode and the“reasoned transparency”mode.Regardless of which concept,doctrine,and model is chosen to fulfill transparency obligation,its content depends on the purpose and use of the academic paper,its system depends on academic norms,and its procedure is for visual justice.
作者
程睿
CHENG Rui(Huazhong University of Science and Technology,Wuhan 430074)
出处
《中国高教研究》
北大核心
2023年第11期62-69,共8页
China Higher Education Research
基金
科学技术部科技创新2030—“新一代人工智能”重大项目“可信人工智能立法制度建设研究”(2022ZD0120100)
2022年度教育部哲学社会科学研究重大专项项目“坚持建设中国特色社会主义法治体系,深入推进全面依法治国实践研究”(2022JZDZ002)的研究成果。