What started as a simple AI-generated ice cream game somehow turned into one of the most weirdly competitive coding experiences I’ve had in a long time.
I joined this ongoing AI coding challenge with extremely low expectations.
Not fake low expectations either. Genuine ones.
At this point, the internet is flooded with AI-generated projects that all somehow look identical after five minutes. Somebody discovers a new model, generates a landing page with glowing buttons and a gradient background, adds glassmorphism cards floating around in space, and suddenly everyone in the comments starts talking like software development has permanently changed overnight.
Most of those projects look impressive right up until you actually interact with them.
Then you realize there’s usually not much underneath.
So when I first opened this challenge, I honestly expected more of the same. A few decent experiments. Some overdesigned dashboards. Maybe a couple of interesting ideas are buried somewhere in the middle.
Then I looked at the leaderboard.
And saw somebody sitting at 96.5.
That immediately changed the feeling of the entire thing.
Because scores like that make it obvious people are no longer treating the challenge casually. Once the leaderboard starts climbing into the 90s, the atmosphere shifts completely. Suddenly, everybody is trying to outdo each other. The projects stop feeling like random experiments and start feeling like actual competition.
And honestly, that’s where the challenge becomes dangerously addictive.
You tell yourself you’re only opening the leaderboard for a minute just to see the newest submissions. Then somebody uploads something absurdly polished, and now your brain immediately starts comparing it to every other project you saw earlier.
A few hours later, you somehow know the names of people climbing the rankings, you’re checking scores multiple times a day, and you’re emotionally invested in strangers competing through browser games and UI animations.
Which sounds ridiculous when typed out like that.
But it’s true.
What surprised me most wasn’t even the AI itself. At this point, everybody already knows AI can generate code, design interfaces, and build basic projects quickly. That part stopped feeling shocking a while ago.
The surprising part was how quickly people started figuring out how to make the projects actually feel good.
That’s a completely different skill.
Because generating something functional is easy now. Generating something memorable is much harder.
And after spending time scrolling through the challenge submissions, that difference becomes painfully obvious almost immediately.
Some projects technically work, but they still feel emotionally empty after a few seconds. You interact with them once, think “okay, cool,” and instantly forget they existed.
Other projects have this weird energy where you can tell somebody is obsessed over the details.
The animations feel smoother.
The interactions feel responsive.
The pacing feels intentional.
The UI reacts properly.
The transitions feel satisfying.
The project stops feeling like “AI-generated output” and starts feeling like an experience somebody cared about shaping.
That gap fascinated me way more than I expected.
Because I think people still underestimate how much personality matters even in AI-assisted development. The projects standing out on the leaderboard usually are not the ones trying hardest to sound technically impressive.
They’re the ones with momentum.
The ones with polish.
The ones with some kind of identity.
That becomes very obvious once you start comparing submissions side by side for long enough.
And honestly, the challenge itself unintentionally exposes something really interesting about the current state of AI creativity.
Everybody participating has access to similar tools.
Most people are using variations of the same models.
Most people have access to similar prompting capabilities.
Yet the results vary massively.
Some entries feel generic instantly.
Others feel crafted.
That difference is not coming from the AI alone.
It’s coming from that direction.
I think that’s why the leaderboard became so interesting to watch after a while. It almost feels less like a coding competition and more like a live demonstration of how differently people think creatively when given the same tools.
Some people focus entirely on visuals.
Download the Medium app
Some focus on interactions.
Some focus on gameplay loops.
Some focus on humor or presentation.
And the really strong submissions usually combine several of those things together in a way that makes the project feel alive instead of merely functional.
That’s probably the biggest shift I noticed while following the challenge.
Not that AI is replacing creativity.
If anything, the challenge made creativity feel more important.
Because once the technical barrier lowers slightly, people immediately start competing through ideas, taste, polish, pacing, responsiveness, and emotional impact instead.
The competition simply moves somewhere else.
And honestly, watching that happen in real time feels much more interesting than most of the usual conversations happening around AI online right now.
Outside spaces like this, AI discussions often become repetitive almost immediately. It’s always the same cycle of arguments about whether AI will replace developers, destroy industries, or eliminate the need for creative work.
Meanwhile, inside this challenge, people are too busy trying to build increasingly polished projects to care about those debates.
That contrast stood out to me constantly while scrolling through the leaderboard.
Nobody there looked passive.
Nobody looked replaced.
If anything, people looked hyper-motivated.
Every few hours, somebody uploads something smoother, cleaner, or more creative than before, and suddenly everyone else wants to improve their own work again.
The challenge accidentally creates this feedback loop where people keep pushing each other creatively without even realizing it.
That’s what makes it difficult to stop checking.
Because the leaderboard never really feels stable.
The quality keeps evolving in real time.
And the higher the scores climb, the more absurd the submissions start becoming.
Some of the projects genuinely do not feel like things built inside a short online challenge anymore. They feel like early versions of products that people could realistically continue developing afterward.
That’s honestly the part that stayed in my head the most after spending time with the challenge.
Not the individual tools.
Not the prompting tricks.
The speed.
The speed at which ideas become interactive experiences now feels completely different from even a year ago.
Not that long ago, most AI-generated projects still had that unmistakable “template” feeling attached to them. You could usually identify them immediately because they all shared the same polished-but-empty aesthetic.
Now the gap between “generated” and “crafted” is becoming much blurrier.
Some entries still feel obviously generic.
But others genuinely have personality.
That feels new.
You can almost watch people learning in real time that successful AI-assisted development has less to do with generating more code and more to do with shaping the experience properly.
That’s why the strongest submissions usually are not the most technically complicated ones.
They’re the ones people actually remember afterward.
And honestly, I think that might end up becoming one of the most important skills in this entire AI era.
Not pure technical execution.
Taste.
The ability to recognize what feels good, what feels memorable, and what makes people emotionally react to something instead of instantly forgetting it.
Because once everybody gains access to powerful generation tools, those softer creative decisions suddenly matter much more than expected.
The leaderboard in this challenge almost feels like a live scoreboard measuring that shift happening in public.
And maybe that’s why the whole thing became unexpectedly addictive to watch.
At first, I thought I was opening another AI coding event.
A few hours later, I realized I was watching people compete creatively in a completely different way than traditional coding competitions usually allow.
Not purely through algorithms or technical difficulty.
But through responsiveness, polish, personality, interaction design, visual feel, pacing, and experimentation.
That combination creates something much more interesting than I expected.
And honestly, I think this is still only the beginning.
If you’re curious, the challenge is still ongoing, and the leaderboard keeps changing. Some of the submissions are genuinely wild to scroll through now:
Because if people are already building projects scoring above 96 in public AI-assisted challenges right now, then the next few years of internet creativity are probably going to get very weird, very quickly.
Top comments (0)