← Ascendy 한국어

meta

Does AI's rise make humans obsolete? — from AlphaGo to Shin Jinseo

· Ascendy Engineering

TL;DR

About this piece. A first-person account pulled out in an interview. “Humans evolve too” is closer to my bet. The Go facts are web-checked (the match was actually 1–1, and the shakier strength comparisons are labeled as estimates in the text). Same vein as Hey, have you even tried it? and I didn’t design it — I delegated it.

I got into coding when AlphaGo beat Lee Sedol

A confession first. I wasn’t originally a developer. What pulled me onto this path was 2016, the AlphaGo vs Lee Sedol match. Just watching, I thought “building something like that would be fun,” and that was the start.

But the message the world read in that match was much heavier. Go had been something like the last sanctuary of human intuition, and it fell. “Humans can no longer beat AI at Go.” Many took it that way, and after that the narrative “as AI advances, humans go obsolete” spread across field after field.

Then this week, the opposite scene arrived

In July 2026, world number one Shin Jinseo won a two-stone handicap game against KataGo, the strongest Go AI in existence. A close fight to move 290, won by 4.5 points.

To be honest, this isn’t a clean-sweep story. Over a two-day match he lost the first game outright and won the second — 1–1. Even so, what makes the win special is which AI it was against. KataGo is the strongest Go AI today, carrying on the lineage AlphaGo opened — the AlphaGo Zero / AlphaZero line of techniques — the lineage of the AlphaGo that toppled Lee Sedol a decade ago. Its algorithms and training are beyond comparison with back then. Against that, a human laid down two stones and won.

At a point where AI has grown this strong, these two stones read as symbolic enough to stand beside Lee Sedol’s single game. (Though with engines so much stronger now, exactly what level “Shin taking two stones from KataGo” meets “the AlphaGo of back then” at is hard to measure precisely — that’s the realm of estimate. But the direction is clear.)

Not “caught up” — the lens changed, and humans evolved

So some people say: “See, humans caught up to AI in the end.” I think that phrasing is wrong. Catching up is a picture of someone behind chasing someone ahead on the same track — but that’s not what happened here.

Over the past decade, pro players studied Go with AI as a teacher and sparring partner. And in doing so, the very way they see Go changed. Moves believed to be settled were overturned; moves humans never played became standard. Shin Jinseo’s Go isn’t a good preservation of the human Go of the past. It’s a new Go, evolved by passing through the lens of AI.

That is, humans didn’t chase AI down and catch it. Inside the new field of view that AI threw open, humans re-evolved themselves. AI evolved, so humans evolved too.

The same for developers — from horses to cars

Here it comes back to my own story. I think this shape applies directly to the developer in the age of AI.

Honestly, the place of the skill I hone has moved. I used to care about being good at debugging, at knowing a language’s syntax precisely. Now — I barely even think about building those abilities. Some developers will call this degeneration. But I call it evolution.

Here’s the analogy. Even after the car was invented, some people still study horsemanship deeply, and that has its own value. But the world’s center of gravity shifted toward driving skill and automotive engineering. What advanced wasn’t riding a horse better, but building and driving a car better.

The developer in the age of AI is like someone who used to ride a horse and hone horsemanship becoming someone who drives a car and hones driving. The place you hone moved from syntax and debugging (horsemanship) to setting direction and judging results (driving). And at the pinnacle is the F1 driver — someone who knows the automotive engineering and drives the best. A seat that masters both the machine and its operation, to the extreme.

So can you know nothing? — a captain doesn’t hold the helm

Here I have to draw a line. “So you delegate everything to AI and a developer can know nothing?” — no. An evolved developer and someone just carried along by AI are different.

I draw the line this way. On a ship, the captain doesn’t hold the helm. A big ship has a helmsman and professional navigators. And yet we say the captain steers the ship. Because the captain makes the judgments that get the ship safely to its destination, and bears responsibility for that voyage.

What separates a driver from a passenger isn’t whether your hands touch the controls. It’s whether you’re making the judgments and holding the destination. A developer who hands implementation to AI but keeps direction and judgment is a captain who doesn’t hold the helm — not a passenger being carried. Conversely, drifting along with whatever AI spits out, with no judgment, is being a passenger no matter how much output appears.

So — “AI’s rise makes humans obsolete” may be a wrong prediction

Ten years ago, before AlphaGo, what we felt was the fear of obsolescence. But Shin Jinseo’s two stones show that prediction is at least not that simple.

Here’s what I want to say. Even to a new tool like AI — and to its almost-impossible-to-adapt-to, staggering rate of advancement — humans are adapting again. As in Go, so in development. When the tool evolves, humans aren’t pushed aside; they re-evolve themselves inside the new field of view the tool opened.

Of course this is my bet. Whether to choose obsolescence or evolution is, in the end, each person’s own. But as someone who started down this path ten years ago because I felt “humans lost” — I rather liked this week’s two stones.

Takeaways

The narrative that “as AI advances, humans go obsolete” may be a prediction we believed too quickly.

Authorship & citation: Written by Ascendy Engineering; quotable with attribution. Found something wrong? Let us know via a GitHub issue.

Tags: ai, go, future-of-work, career, evolution, opinion