Why we still build, even when machines can do it faster.

I love scribbling down thoughts. There is a specific joy in taking difficult, complex concepts and breaking them down into soft, digestible pieces that anyone can understand.

I also enjoy drawing. Although my skills are strictly "programmer art" level, I believe a single clean diagram is often far more powerful than a hundred words.

And above all, I love coding. I truly cherish the process of creating something that actually works right at my fingertips.

I first dipped my toes into the massive wave of Generative AI back in 2023, right when Stable Diffusion was starting to take off. I had poked around game AI before that, but looking back, the shift that began then seems to be shaking the entire IT ecosystem to its roots.

At the time, I felt a strange thirst when looking at models trained primarily on Western art styles. So, I spent time painstakingly crafting datasets and fell deeply into the fun of teaching AI the brushstrokes of Shin Yun-bok, a painter from the Joseon Dynasty.

Then came a period where I felt infinitely small in front of the high-quality images pouring out so effortlessly. What sustained me through that overwhelming feeling was the realization that "teaching a new style and setting the direction" was still a human task.

A similar sense of helplessness arrived with writing. Watching models evolve from ChatGPT and Gemini, I witnessed AI’s writing skills quickly surpass my own. However, I realized that deciding what to write, bearing the weight of a piece published under my name, and finally putting the period at the end of the sentence is something only "I" can do. That sense of responsibility is something AI cannot take away.

Is coding any different? Although I still do a lot of the typing myself, my jaw drops at the speed of evolution every time a new model is released.

When it comes to writing or drawing, AI is already the superior player. So, a collaborative process has settled into my daily life: I throw out a rough draft, the AI polishes it smoothly, and I do the final review and adjustments. As the models get smarter, the parts I touch are becoming fewer and fewer. I have a hunch that coding will follow this exact process before long.

At an AI Workshop I attended yesterday, someone asked me a heavy question:

"What on earth should humans do in the future?"

As is the case now, even more things will be automated by AI in the future.

However, people like me will still want to make things ourselves. Even if we borrow the power of a tool as potent as AI, the starting point and the intention of that creation will still remain with the "person."

There will be a clear distinction between what AI generates because it "wants" to (if ever), and what a human creates with specific intent. The value will likely be assessed differently, too.

Isn’t it similar to the variety of choices we have when we need a chair?

Sure, you can pay money and buy a comfortable, finished product. But some enjoy the process of buying parts from IKEA and assembling them; others buy the tools and cut the lumber to build from scratch; and some even choose the primitive labor of carving wood, stitch by stitch, by hand.

Just as we pay different prices for factory-made goods and artisanal handicrafts today, I believe the "us" of the future will continue to live on, assigning different meanings based on the "process" and "values" embedded in the result.