The Future of Work in the Age of AI: Lessons from Seoul

Made by : Marla, Pato, Sofia and Mariana

A city can feel futuristic. A classroom can make it feel human again.

Magazine feature
Reporting and reflection from Seoul, class discussion, and today's talk.

Seoul does not introduce technology with a drumroll. It just appears: in classrooms, in services, in the rhythm of daily life. Since arriving here, I have noticed how naturally technology is integrated into everyday routines. At first, that felt intimidating. After yesterday's class and today's talk, it feels more complicated — and more interesting.

AI is changing work, but not every job in the same way. Some tasks are exposed because they are repetitive, text-based, predictable, or easy to evaluate. Others remain deeply human because they depend on trust, safety, field judgment, communication, and the ability to care about another person.

“Technology rarely destroys a profession. It transforms it.”Quote from today's talk

What yesterday's class made clearer

Yesterday, my professor explained that AI is trained on enormous amounts of text — he mentioned around 13 trillion words, although technically it is trained on tokens. The point was not that AI “knows everything.” It was almost the opposite. He emphasized that AI is not actually as intelligent as many people believe and that it does not simply store everyone's personal information. Instead, it predicts patterns based on its training data.

That matters because people often talk about AI as if it were a mind. In class we also discussed that AI often tells users what they want to hear. My opinion is that, since AI was created by humans, it will always need people to guide, improve, and supervise it. Rather than seeing AI only as a threat, I think it should be viewed as a powerful tool.

Still, there is a worry I cannot ignore. AI is already capable of performing many tasks that people spend years learning, such as design or architecture. I do not believe that automatically means humans become obsolete. Perhaps AI is not advancing faster than humanity — perhaps we are simply relying on it too much.

Classroom detail13 trillion words versus tokens.

Personal opinionAI needs human guidance and supervision.

Seoul observationTechnology feels naturally integrated into daily life.

ReflectionThe danger may be dependence, not just AI itself.

Talk connectionEducation is transforming, not ending.

ConclusionAI works best as a tool, not a replacement.

At Creverse, the lesson was not just speed

During today's talk, someone asked how AI had affected employment at Creverse. The speaker explained that in Seoul there has not been a major risk of teachers losing their jobs because their work is centered on mentoring and supporting students rather than only grading assignments. He contrasted this with countries where automation is creating greater employment uncertainty.

Creverse replaced many grading tasks with an AI system called Hummingbird. After only three months, operational costs were reduced by more than 80%, feedback time fell from 2–5 days to approximately 5 seconds, and revision behavior changed dramatically.

80%+operational cost reduction

5 secapproximate feedback time

2–5 daysprevious feedback wait

25%students revised before AI

The most interesting part is not the number. It is the behavior behind the number. When feedback arrives days later, the student has already moved on emotionally. The assignment feels finished. But when feedback appears almost instantly, revision becomes part of the learning moment itself. Students are not just receiving a grade; they are being nudged to try again while the idea is still alive.

“AI can grade essays in seconds. Humans still inspire students.”Quote from today's talk

Which jobs are most exposed?

AI does not hit “jobs” as a whole. It hits tasks. The Anthropic Economic Index studies AI's effects on labor markets by looking at real AI interactions, while McKinsey's recent work on AI trust argues that organizations still face gaps in governance, risk management, and accountability as AI becomes more agentic. GitHub's Copilot research gives a concrete example: developers using Copilot completed a programming task substantially faster in a controlled experiment.

Jobs Most Impacted by AI Automation

The chart shows why essay grading is more vulnerable than civil engineering. Grading follows patterns: rubric, text, feedback, score. Civil engineering still depends on human judgment in the physical world. Engineers must visit sites, understand safety, communicate with teams, negotiate constraints, and make decisions when conditions are messy. A model can support calculations or produce drafts, but it cannot take responsibility for a bridge, a road, a mistake in the soil, or the people who will use the structure.

Comparison: What AI can automate vs. what people still protect

Work area	AI strength	Human strength
Essay grading	Fast pattern-based feedback and rubric checking.	Motivating students and understanding emotional context.
Software development	Code suggestions, boilerplate, debugging support.	Product judgment, responsibility, ethics, and collaboration.
Design and architecture	Rapid concepts and visual alternatives.	Taste, purpose, client needs, culture, and lived experience.
Civil engineering	Analysis support, simulations, drafting assistance.	Safety decisions, field work, communication, and accountability.
Teaching	Instant feedback and practice support.	Mentoring, inspiration, trust, and care.

Photography, painting, and the fear of replacement

One quote from today's talk stayed with me because it made AI feel less like a brand-new panic and more like part of a pattern.

“Photography didn't kill painting. It helped painting redefine its role.”Quote from today's talk

1838

Photography invented
Reality could be captured mechanically.

↓

Painting evolves
If a camera could capture reality, painting could move deeper into imagination.

2026

AI transforms education and work
Routine tasks become faster, cheaper, and more automated.

↓

Human work evolves
Creativity, judgment, empathy, and responsibility become more visible.

As the speaker put it: “Photography captured reality. Painting discovered imagination.” Maybe AI will force a similar question. If machines can produce answers, what should schools, teachers, workers, and students become better at?

The human part is not decoration

One of the most dramatic quotes from the talk was from Jimmy Carr: “The fourth humiliation is AI. We're now not even the smartest thing.” Another line asked: “It's all-knowing. All-powerful. Can create miracles. Lives in a cloud. God or AI?” Those quotes are funny, but also uncomfortable, because they describe how easily people can exaggerate what AI is.

The more practical questions were sharper: “Will AI kill learning centers?” and “Who even needs schools? Who needs colleges in the coming age of AI?” The answer I took from Seoul is not that schools disappear. It is that schools have to prove what only humans can do.

Since arriving in Seoul, I have noticed how naturally technology is integrated into everyday life. Initially that felt intimidating, but after everything we discussed in class and during today's talk, I no longer see AI as something inherently dangerous. Instead, I see it as a tool that can help people become more efficient while still requiring human creativity, judgment, and critical thinking.

AI does not end education. It forces education to evolve. It does not erase work. It changes what parts of work deserve our attention. The future belongs less to people who pretend AI does not exist, and more to people who know when to use it — and when to disagree with it.

External evidence used

The article uses the classroom and talk content above for personal experience. External sources were used only for broader evidence about AI, work, productivity, adoption, and governance.

Anthropic Economic Index — initiative studying AI's effects on labor markets and economy through usage data: anthropic.com/economic-index
GitHub Copilot productivity research — controlled experiment reporting faster task completion with Copilot: github.blog
McKinsey State of AI trust in 2026 — trust, governance, risk management, and the agentic era: mckinsey.com
McKinsey State of AI 2025 PDF — AI agents, innovation, and organizational adoption: mckinsey.com PDF
Innopharma Education — AI changing job roles, workforce needs, and upskilling expectations: innopharmaeducation.com