meta
Hey, have you even tried it? — the fastest way to get good at AI
· Ascendy Engineering
TL;DR
- If I had to play Go against AlphaGo in a month, I’d ask AI “how do I play Go well?” If I had to build AlphaGo in a month? The questions change completely. A month later, who handles AI better — obviously the builder.
- The fastest way to get good at AI is boring: use it a lot. But using it a lot only to search and ask won’t get you there. It’s not about volume — it’s about mode. Build something.
- Don’t be afraid. Even on a problem a junior can’t solve, it isn’t whether you solved it — the process of thinking, trying, and failing is the asset. That’s compressed real-world experience.
- There’s one real fork: talking about it in theory and words, or — as Chung Ju-yung, founder of Hyundai, put it — “Hey, have you even tried it?”
About this piece. A first-person account pulled out in an interview. “Build something” is my bet, not a verdict. Same vein as to learn a domain, build with AI and what should you study in the age of AI — but this one is centered on getting good at using AI itself, fast.
In a month, will you play AlphaGo or build it?
Start with a thought experiment. Say I have to play Go against AlphaGo in a month. I’d probably ask AI something like: “How do I play Go well? How do I beat AlphaGo?”
But say I have to build AlphaGo in a month. The questions I ask change entirely. Which search algorithm, how to run the training, where it breaks down… the whole layer of the questions drops down.
And a month later, which of the two handles AI better? Obviously the builder. That difference is the whole point of this piece.
The boring answer — that’s only half right
Ask “what’s the fastest way to get good at AI,” and the answer is honestly boring. Use it a lot. That’s true; there’s no shortcut to getting better.
But this is where most people hear only half. If you take use it a lot to mean using it a lot as a search tool or a question box, that’s wrong. Used that way, you won’t improve much no matter how much you use it. It’s not a volume problem — it’s a mode problem. Consuming answers by asking and creating by building are completely different.
That’s what the AlphaGo analogy is about. The user and the maker start from a different vantage point. So they ask different questions, and those different questions are what raise your AI skill.
So — build something
My recommendation is simple. Build something. And specifically, use AI to build a tool that solves a problem in a field you care about and want to work in.
Here comes the common objection. “I’m a junior — how would I know what the problems are? These are problems seniors couldn’t solve for years. Will just using AI really crack them?”
Fair. It probably won’t get solved. But this is where you have to flip the frame.
The flip: the process, not the result, is the asset
You don’t only learn by solving the problem. If anything, learning and growth happen in the process of spending your time thinking, trying, and failing — the process where that data accumulates.
Hard problems are fine to fail. The experience of having tried to solve one, the time you put in, your own wrestling — that alone becomes your asset. Juniors are indeed the most exposed in the age of AI. But if a junior can stack up this kind of compressed real-world experience, there’s no reason to be afraid.
Where to start — a field you care about
So why specifically a field you care about and want to work in? It’s not only about motivation.
You need to know something even to ask a question. In a field where you know nothing, even the first question stalls. But in a field you’re interested in and want to work in — even if you’re close to a total beginner — you likely know at least one thing. That’s why the interest is there in the first place. And that one thing becomes the foothold to start asking anything at all.
The real fork: have you even tried it?
By now you might think not all building is equal building. True. But the real fork I felt was before the question of what to build.
Honestly, I too used to not actually build anything. I thought in theory, and talked about it in words and writing. Then I actually put my hands on it and built — and the AI I experienced was an entirely different thing. The AI I imagined in words and the AI I hit while building were not the same.
Among my peers, too, plenty just talk. Every time I see them, one line rises to my throat — what Chung Ju-yung, the founder of Hyundai, reportedly threw at executives who hesitated before a hard task:
“Hey, have you even tried it?” (이봐, 해봤어?)
The most honest thing I can tell someone who wants to get good at AI fast is exactly this. Before you take another course, before you search for the perfect prompt — have you tried it? Have you built even one small thing?
Coda: what remains, win or lose
I’m building a product right now. Whether it works out, I don’t know. But one thing is certain — win or lose, what remains for me is the experience and know-how of how to use AI, how to build a harness that improves the work, and the domain knowledge of my field.
That’s the heart of it. The real reward of building isn’t the artifact — it’s what accumulates inside you while building. And that doesn’t disappear even if you fail.
Takeaways
- Use it a lot — but while building. Using it a lot only to search and ask won’t get you there. It’s mode, not volume.
- The user and the maker ask different questions. Playing AlphaGo vs building it — a month later, the builder handles AI better.
- The process, not the result, is the asset. Even unsolved, the accumulation of trying, time, and wrestling becomes compressed real-world experience.
- Start in a field you care about. You need to know something to ask; that field is your foothold for the first question.
- And — have you even tried it? Not theory, words, or writing, but whether you’ve built one small thing yourself: that’s the fork.
Honestly this too is my bet, not a verdict. But if you want to get good at AI fast, before piling up more courses and prompts, step over to the maker’s seat — that was the fastest path I found, coming all this way from a non-specialist start.
Authorship & citation: Written by Ascendy Engineering; quotable with attribution. Found something wrong? Let us know via a GitHub issue.
Tags: ai, learning, career, building, opinion