Abstract: Rules of transparency are the basic principles of trustworthy AI, and obligation of transparency is the responsibility mechanism for designers, developers, and users of AI to achieve visual justice through technological and legal due process. As a transition from artificial narrow intelligence to artificial general intelligence, ChatGPT has the ability to generate structured content similar to that created by human. While improving academic capabilities, it also poses risks to academic integrity. ChatGPT's intervention in the creation of academic papers involves the underlying logic of copyright, the standard logic of academic integrity, and the top-level logic of trustworthy artificial intelligence. All three logics require authors to fulfill their obligation of transparency. Obligation of transparency is the proof of the performance of digital obligation, the display of digital fairness, and the expression of digital responsibility. It has the guarantee function of academic integrity, academic responsibility, and academic fairness.

摘要： 透明度规则是可信人工智能的基本准则，透明度义务是人工智能的设计者、开发者、使用者通过技术性正当程序和法律正当程序实现可视正义的责任机制。作为弱人工智能向强人工智能过渡的ChatGPT具有生成类似于人类创作的结构化内容的能力，其在提高学术能力的同时也会对学术诚信带来风险。ChatGPT介入学术论文创作涉及到著作权的底层逻辑、学术诚信的本位逻辑、可信人工智能的顶层逻辑，三重逻辑都要求作者履行透明度义务。透明度义务是数字义务履行的证明方式、数字公平的展示方式、数字责任的表达方式，具有学术诚信、学术责任和学术公平的保证功能。