[Class Report] Introduction to System Development, Week 24
Topic: Introduction to Generative AI – Basics of Mechanism and Usage
In Week 24, in preparation for next week’s practical exercises on Generative AI, we had a preliminary session covering basic concepts, ethics, and introductory prompt design. Building on what we learned about external API integration, the class explored the question: “What can generative models actually do?”
■ Teacher’s Introduction: “Know the tool, and learn to use it correctly first”
Mr. Tanaka:
“Generative AI is a very powerful tool. But first, we must understand its mechanism, what it can and cannot do, and the rules for using it.”
He began with a simple analogy: Generative AI is like “a giant warehouse of text that assembles new sentences by referencing existing ones.” He stressed that, unlike humans, it is not always correct.
■ Today’s Key Points: Fundamentals and Cautions in Generative AI
The lecture summarized the following items:
- What is Generative AI?: A general term for AI that generates text, images, or audio (e.g., text models, image models).
- Training data and bias: Since models learn from past data, bias may be embedded.
- Limits of accuracy: Outputs are not reliable primary sources—fact-checking is essential.
- Ethics and copyright: Generated content may raise copyright or data source issues.
- Privacy and safety: Avoid inputting personal data; keep API keys secure.
Student A: “I didn’t realize AI could be wrong—that’s surprising.”
Student B: “If the data is biased, that could be pretty scary.”
■ Exercise ①: Introduction to Prompt Design
The first practical step in “getting useful answers from AI” is prompt design. Students compared poor vs. improved prompts.
Poor Prompt Example
“Tell me about travel.”
Improved Prompt Example
“Please suggest 3 nature spots near the Kanto region that high school students can visit for a one-day weekend trip, considering train access and a budget under 2000 yen. List them briefly in bullet points.”
The teacher explained the importance of being specific, giving constraints, and specifying the output format.
Student C: “Just asking for bullet points makes the answer way easier to read.”
■ Exercise ②: Mini Prompt Workshop
Each group was given a short task on paper and asked to design a “good prompt.” Example themes:
- Quick bento menus (constraints: no allergies, max 3 ingredients)
- Quiz questions for class (constraints: 3 questions, with correct answer + explanation)
- A draft guide for a book report (constraints: for high school students, max 400 characters)
Groups specified who the answer was for, what the task was, and in what format the response should be. They then shared prompts with the class.
Student D: “Adding constraints makes the answers much more practical!”
■ Mini Lecture: Ethics and Practical Rules for Using Generative AI
The second half of class was a short lecture on the “rules to follow” when using AI:
- Always verify sources: Check facts with another source.
- Respect copyright: Do not copy or adapt specific works directly.
- Protect personal info: Don’t input names, contacts, etc.
- Follow school rules and service terms.
- Transparency: If you use AI in assignments, disclose it.
Mr. Tanaka:
“The more convenient the tool, the more important it is to follow the rules. Misuse could harm others or backfire on yourself.”
■ Teacher’s Note
“Generative AI is strong at idea generation and reducing repetitive work. But the goal is not to leave everything to AI—it’s to collaborate with AI to create something good. Learn to use it correctly first.”
■ Homework (Short Practical Task)
To prepare for next week’s hands-on session, students received the following homework:
- Write one “improved prompt” at home (any theme).
- Write a sample expected output (50–100 characters).
- List up to 3 points you considered when creating the prompt.
The teacher advised: “Actually writing prompts yourself is the fastest way to improve.”
■ Next Week’s Preview: Hands-On with Generative AI
Next week, students will actually try prompts in a safe classroom environment (with a sandbox or pre-prepared materials). Activities include:
- Running basic prompts
- Verifying outputs
- Practicing output formatting (e.g., bullet points)
In Week 24, students learned both the potential and risks of generative AI. They left class realizing it’s both “useful” and “a responsibility,” and showed strong motivation for next week’s hands-on session.