May 26, 2023
The University of Tokyo has provisionally established the following principles regarding the use of ChatGPT and other generative AI tools in educational settings. We encourage awareness of the strengths and weaknesses of generative AI tools and their appropriate use.
- University’s Position: The University of Tokyo does not enforce a blanket prohibition on the use of generative AI tools. Instead, we are actively exploring their potential and plan to provide practical guidance for their use.
- Use in Classes: The decision whether or not to allow the use of generative AI tools will depend on the class. This decision, along with the conditions for their use, is left to the discretion of individual instructors. Please follow the guidelines set by your department, section, and each class instructor regarding the use of these tools.
- Educational Impact: University education places a premium on enhancing students’ thinking abilities through the process of acquiring and refining knowledge. Generative AI tools automate the collection and organization of information and display only the results. Relying solely on generative AI tools for class assignments could deprive students of essential opportunities to exercise their thought processes, potentially impeding long-term personal development. This is why certain classes may prohibit the use of these tools or set specific conditions for their use.
- Academic Misconduct: Respecting each person’s ideas and originality is paramount not only in academia but in all social activities. When producing reports and papers, it is essential to properly cite sources and express your unique thoughts. It is unacceptable to present text generated by AI tools as your own when submitting class assignments.
- Potential for Errors: Text generated by AI tools may contain errors or information that does not align with the user’s intentions, even if it seems plausible at first glance. AI responses can also contain biases based on their training data and algorithms. Rather than accepting AI-generated results at face value, it is crucial to scrutinize them and make adjustments as necessary. As with internet searching, assessing the credibility of AI-generated content and making corrections is a vital skill in today’s society.
- Legal Risks: It has been suggested that the output of AI tools, not just generative ones, may potentially pose problems related to copyright and design rights. Therefore, it is important to note that using generated results as is may entail legal risks in the future. In some cases, it may be necessary to indicate which tools were used.
- Information Security: Information entered into AI tools can potentially be used in some form for AI training. Never input confidential information, personal information, or unpublished research results, as the information might be leaked.
- Future Plans: AI tools are constantly evolving, and the precautions for their use may change accordingly. The above guidelines are based on the situation as of May 2023. We will continue to update them as necessary in the future.
Kunihiro Ohta, Ph.D.
Executive Vice President of the University of Tokyo in charge of education and information technology