This is a review of an on-site training session for the Korea Foundation for Labor-Management Development.
How are people who use AI tools well different?
Takeaways after the lecture: “Start by imagining the result” really is the key in the AI era.
Today at the Seoul Technical Education Center, Jungbu Campus in Hannam-dong, I gave a special lecture on “How to Use AI Tools” to trainees in the Software Testing Expert Training Program.
With AI now a working tool, it was a chance to share, in practical terms, how to use it well.
In particular, I worked in an episode from mid-lecture, when I got a call requesting consumer trend training, and walked through it as a live example. It drew people in.
Designing an AI tool workflow for a real-world scenario. "People who use AI well imagine the result first, then design the steps to get there."
This was the key phrase of today's special lecture.
For example, during a break I got an urgent request for a 2026 consumer trends training session.
In that moment, I pictured the end result first.
"A finished slide deck + trend categories + the latest data." With that end result in mind, I broke the work down like this.
Step 1: Research. Step 2: Content summary. Step 3: Slide material. Step 4: Error review and fixes. Step 5: Presentation flow. Then match an AI tool to each step of the workflow.
The judgment to decide what to hand to AI. Deciding what people should do and what AI can do is still up to humans.
Right there, I searched keywords related to 2026 consumer trends on my phone and fed them into an AI summarization tool to pull out the key points.
Next, I asked an image-generation AI what style the slide visuals should take. Finally, I checked whether "this flow feels natural for a presentation," polishing the wording and settling the tone.
Keep AI working even while you’re on the move. That’s the real key.
You should be able to keep AI working around the clock, even when you're on the move or no one is at their desk.
For example, you can have AI draft your slides from your phone, open the share link to check for errors along the way, queue up image requests, and get a push notification when they’re done, so all you do is pick the finished images.
How you write the prompt is 80% of the result. More often than not, the problem isn't that AI can't do it. It's that we couldn't describe exactly what we wanted.
One of the prompts I used this time went like this.
"Summarize the 2026 consumer trends 'value-driven consumption,' 'time-saving consumption,' and 'Gen Z-led trends,' and include sources.
Summarize it in no more than 2 lines per sentence, so it works well on a slide." A precise, specific request can boost AI performance tenfold.
Where must a human step in?
Handing work to AI is convenient, but it can’t yet be trusted with everything.
So in the session, we also practiced defining the "human role" using a table like this one.
Step / Role / Owner / Task. Step 1: Human, set direction and imagine the result. Step 2: AI, collect and summarize data. Step 3: AI, visualize and generate a draft. Step 4: Human, review sentences and adjust flow. Step 5: Human, rehearse the presentation and prepare for Q&A. The reality of building a live lecture: between fun and pressure. Honestly, AI lectures are changing incredibly fast.
A deck you build today can be out of date within a week.
Building new course material for every case can be a burden. Still, creating practical content people can use on the job right away may be a core skill instructors can’t afford to miss today.
More than the question itself
A lecture built on hands-on experience. The biggest takeaway from this session was that participants felt “AI is a tool” not just in their heads, but at their fingertips.
If this lecture was more than a tool introduction, and you left thinking “I really should try applying this to my own work,” then that was today’s goal, and I think that’s why I keep doing this work.
