2024.03.12 - コミュニケーション学科  AI and Making an English Test

Have you had the chance to explore AI ChatGPT? We utilized this technology to create one of the English tests. The experience? It didn’t necessarily reduce the time spent creating the test; instead, it altered the allocation of time to different aspects of test-making.
For instance, AI ChatGPT proved invaluable in quickly producing and editing texts. However, when it came to generating accurate and suitable test questions for the student’s level, it demanded more time and attention from us. It often gave us the wrong answers to the questions as well! At this stage, it felt like we were collaborating with a child. In its defense, we were like children regarding prompts. In other words, we needed to learn how to better communicate with it. This common communication issue between humans and AI has led to a new profession called ‘prompt engineering.’
A ‘prompt engineer’ possesses natural language expertise in communicating with large language models (LLMs), such as ChatGPT. LLMs are AI systems capable of understanding and generating language. One widely adopted prompt framework is called RISEN:
• Role: Define its role, for instance, as an expert English test writer.
• Instruction: Provide clear steps with a specific goal.
• Steps: Outline the necessary actions to achieve the goal.
• End Goal: Clearly state the final objective.
• Narrowing: Establish constraints, like limiting the test to 20 questions.
It is fascinating that we must develop new communicative skills to work with AI! I hope you found this column interesting.