Is Generative AI a Magic Secretary? How Work Changes: Usage and Precautions for Business Beginners

Introduction: Generative AI is Actually an "Excellent New Employee"

Recently, you've likely heard the term "Generative AI" frequently in news or colleague conversations. Honestly, haven't you felt, "It seems complicated" or "Only engineers can use it, right?"
Please rest assured. Generative AI is not a robot calculating complex formulas; it is a tool that understands human language and creates new text or images.

If we compare it to a corporate organization, Generative AI is like an excellent new intern who is skilled in languages, reads books quickly, and can swiftly draft materials in any field. Of course, they lack experience and may misunderstand things occasionally, but if you give clear instructions, they become a strong partner that significantly saves your time.

This time, we will explain how to work with this "new intern" clearly, avoiding jargon.

1. Document Creation & Summarization: A "Supporter" for Tedious Admin Tasks

Concept Explanation: An Analogy Using Cooking?

What Generative AI excels at most is generating text. If we compare this to cooking, it acts like a chef who prepares ingredients and plates the dish for you just by receiving your order, such as 'Japanese-style Pasta, Salt Flavor'.

You don't need to make the "finished high-end dish" yourself. By thinking about "what kind of menu flavor you want" and leaving the cooking to the AI, you save time.

Specifically, this applies to polite thank-you emails for external parties, summarizing meeting minutes from notes taken during meetings, and organizing key points of long reports.

Business Use Cases

For example, when creating a customer complaint response email, you typically type while thinking about "Apology," "Explanation of Cause," "Prevention Measures," etc. However, if you instruct the AI as follows, a draft is created instantly.

"I want to create an apology email to a client. The content should apologize for the inconvenience caused by delays and convey that we will implement system redundancy as an improvement measure. Please use polite honorifics."

[Before/After]

• Before: Takes 30 minutes to think about the structure, choose words, and revise in my own head.
• After: Input the instruction and get a draft in 30 seconds, then just minor adjustments. Total time 5 minutes.

2. Planning & Idea Generation: Utilizing as a Sparring Partner

Concept Explanation: An Analogy Using Company Organization?

When you can't decide on a new product catchphrase or can't come up with event ideas, idea generation is effective. This is like having a brainstorming partner who is always available and positively accepts everything.

With humans, you have to be considerate or worry about silence, but with AI, you can push back repeatedly asking, "Isn't there a more interesting idea?"

Business Use Cases

For marketing personnel, copywriting targeting the audience can often be a headache. By setting a "persona (assumed customer)" for the AI, you can draw out concrete proposals.

"Imagine you are a fashion leader in your 20s. Think of 5 promotional slogans for spring's new one-piece dresses that would interest them. Include a bit of humor."

[Before/After]

• Before: Struggling alone yields no ideas, leading to mental stagnation all day.
• After: Diverse angle proposals are presented immediately, allowing you to focus only on selecting and combining good elements.

3. Data Analysis & Programming: A "Translator" Anyone Can Use Without STEM Knowledge

Concept Explanation: An Analogy Using Cooking?

To those struggling because "I don't know Excel functions" or "I can't write programming code." Generative AI also serves as a translator who translates technical jargon into Japanese.

In cooking terms, it's like a recipe site that tells you, "If you simmer this ingredient at this heat level, it turns out like this." Even non-experts can achieve results simply by following the steps.

Business Use Cases

If a salesperson wants to extract data matching specific conditions from a customer list (CSV file), they don't need to be good at Excel's VLOOKUP function. Enter the following into the AI, and it will provide the necessary formula.

"There is an Excel table with names in Column A and ages in Column B. Create a formula to filter only people aged 30 and older."

Alternatively, it is possible to generate the code for a simple website form itself and ask an engineer to review it.

4. Customer Support & FAQ: A "Front Desk" Operating 24 Hours

Concept Explanation: An Analogy Using Company Organization?

Customer inquiry handling is a task that easily exhausts staff because it requires answering the same questions repeatedly. Here, we position AI as a concierge who has mastered the manual perfectly and works 24 hours a day, 365 days a year without rest.

However, this is strictly for "initial reception" or "standardized responses." Handling customers with heightened emotions ultimately requires human intervention.

Business Use Cases

Used for updating content on Frequently Asked Questions (FAQ) pages or setting chatbot responses.

"Write an article for the FAQ page regarding our company's credit card annual fee waiver conditions. The key points are 'Annual Usage Count' and 'Usage Amount'."

[Before/After]

• Before: The person in charge spends 1 hour per week updating manuals and creating answer templates.
• After: When conditions change, issue a correction instruction to the AI and reflect it immediately. Zero man-hours.

5. Risk Management: "Safe Driving Rules" for Using AI

Concept Explanation: An Analogy Using Cooking?

Convenient tools always carry risks. This is the same as requiring heating when handling raw meat and avoiding allergenic ingredients.

AI sometimes lies (this is called hallucination). There is also a possibility that input information could leak externally. Creating rules to prevent these is essential.

Business Use Cases

Security Measures: Important customer names, passwords, and corporate confidential information must never be input into AI. Before inputting, processing such as masking personal information (e.g., Mr./Ms. XX) is required.
Information Verification: It is an iron rule to always verify facts regarding numbers or legal interpretations generated by AI.

Q&A Answering Common Doubts

Q: What is the difference between the paid and free versions?
A: Basic text creation is possible with the free version, but the paid version often has faster processing speeds or quicker access to latest information. Trying the free version first is recommended.

Q: Can't I use it if I only speak English?
A: No problem at all. If you give instructions in Japanese, it responds in Japanese. Conversely, it is possible to read English materials and summarize them in Japanese.

Q: Can I use the generated text as is?
A: Basically, treat it as a "draft." Check it with your own eyes once for AI expression quirks or typos before using it.

Where to Start? Concrete First Steps

Something you can start doing immediately today is "Drafting a morning meeting email". Instead of writing it yourself as usual, try asking ChatGPT or Microsoft Copilot to "Please propose an agenda for tomorrow's regular meeting."

Even just fixing the proposal born there slightly, you should feel a completely different "sense of speed" compared to before. Let's start with small experiments first.

Glossary: 5 Keywords You Should Know

• Generative AI: Artificial intelligence that creates new content such as text or images.
• Prompt: Words used to command or instruct the AI (like a spell).
• LLM (Large Language Model): The brain of Generative AI, a mechanism trained on massive amounts of data.
• Hallucination: The phenomenon where AI outputs plausible-sounding false information (illusions).
• Copilot: The name of an AI service that can integrate with office software, provided by Microsoft, etc.