[Cha Seunghui's AI Redesign Strategy] ⑫ The Art of Delegating a Draft

A Four-Step Structure for Delegating Drafts

[Image Source = NotebookLM]

[Korea Lecturer News Columnist Cha Seunghui] You have your raw material. The structure is in place. Now it is time to hand the draft off to AI. But when you actually do, the result often falls short of what you expected. The direction is off, the content is too generic, or the writing style feels nothing like your own.

This is not a limitation of AI. It is a problem with how the delegation was handled. Delegating a draft well is itself a skill.

■ Delegation and Abandonment Are Not the Same Thing

When most people hand off a draft, they do it like this: throw a topic at AI and wait for results. Then when the output disappoints, they blame the tool.

But think about it. When you assign a report to a team member, do you simply hand them a topic and wait? Of course not. You explain the purpose, identify the audience, specify the length and format, and highlight the key points that need to be addressed. Delegating a draft to AI is no different.

Delegation is not about throwing something over the wall. It is about providing enough information to set the right direction. Delegation and abandonment are entirely different acts.

■ A Four-Step Structure for Delegating Drafts

The field-tested approach to delegating drafts follows four steps.

Step 1: Start with the purpose and the audience. Communicate why the draft is needed and who will read it. Be specific — "This is for a report to my team leader" or "This is aimed at office workers who are new to AI." When the purpose and audience are clear, AI selects the appropriate language and depth accordingly.

Step 2: Agree on the structure first. Before any content is written, confirm the framework. Ask AI to review your intended structure — "I'm planning to write with this outline. Please check whether anything is missing." If content gets filled into a flawed structure, the whole thing may need to be rewritten. Getting the framework right first narrows the scope of revision later.

Step 3: Specify the length and format. Be concrete: "Fit it on one A4 page," "Organize it under three subheadings," "Keep each section to two paragraphs or fewer." Without a format, AI will make its own judgment calls to fill the space. When those calls differ from your expectations, unnecessary revisions pile up.

Step 4: State what to leave out. What you do not want matters just as much as what you do. Conditions like "avoid overly academic language," "start with the problem rather than the conclusion," or "do not use a list-based structure" give AI clear boundaries to work within — and the result lands much closer to what you had in mind.

People who delegate well to AI know how to articulate both what they want and what they do not want.

■ Do Not Try to Get It Right in One Shot

The most common mistake in delegating drafts is expecting a finished product from the first attempt. A draft is, by definition, a draft. There is still work ahead — confirming the direction, filling in the gaps, and rewriting in your own voice.

Demanding completion upfront actually makes the result worse. AI will pad the content or fill it with vague language to make it look finished. The output may seem polished on the surface, but there is often little in it that is actually usable.

Instead, break the process into stages. First confirm the structure, then fill in the content for each section, and finally refine the language. At each stage, your own judgment enters the process — and the result gradually becomes more truly yours.

■ A Draft Is Not a Deliverable — It Is the Start of a Conversation

No one uses a draft as-is. What matters is how you work with it.

When you treat a draft as the start of a conversation rather than an end product, your approach changes entirely. You check whether the direction is right, identify what is lacking, and ask for something better. Through this process, the dialogue with AI becomes increasingly precise — and so does the quality of the result.

More important than receiving a good draft is knowing how to use one. And the key to that lies in what the next issue covers: the art of revision.

▶ Next Issue Preview

Next time, we explore "The Art of Revising a Draft." There is a significant difference between using an AI-generated draft as-is and revising it properly. We will look at what to change, how to change it, and the specific criteria that turn a draft into something truly your own.

Columnist Profile

Cha Seunghui is an AI strategy expert specializing in the redesign of work and learning through AI. Going beyond simple technology adoption, she champions practical AI education that transforms the way people think and structure their work. She continues to lecture and conduct research across public institutions, universities, and corporations in the areas of AI utilization strategy, workplace innovation, and digital literacy.

She serves as an Adjunct Professor in the Department of Computer Engineering at Yeungnam University, and is actively engaged in educational settings as a Security-Designated Instructor for the Ministry of National Defense, an AI Instructor at Korea University of Technology and Education (KOREATECH), an AI Instructor at Line Distance Lifelong Education Institute, and a Supervising Professor at the Digital Convergence Education Institute. She also contributes to the growth of the AI industry and education ecosystem as a Research Fellow at the AI Election Strategy Research Institute, Senior Researcher and Daegu Branch Director at the GEO Marketing Research Institute, Vice President of the Korea AI Video Production Association and the Korea AI Literacy Instructor Association, and Director and Gyeongsang Chapter President of the Korea AINFT Association. She previously conducted research in intelligent robotics at Daegu Gyeongbuk Institute of Science and Technology (DGIST) and currently serves as Chief Researcher at Near Networks Co., Ltd.

Her published works include Theory of Programming Languages, Workplace Innovation Using Generative AI, This Is GEO Marketing, AI-Powered Election Strategy, The Age of Being Recommended by GEO AI, How AI Agents Are Transforming Public Administration, and The Future Shaped by Physical AI and Agentic AI. Throughout her career, she has offered actionable insights that bridge technology and strategy. As a certified AI Education Instructor and Prompt Engineer, she is actively working to expand practical AI literacy in real-world settings.

주요

기사

Columnist Cha Seunghui