In Thai, calling someone a "duck" is a sarcastic comparison. It can fly but not well, swim but not fast, run but not far. A jack of all trades, master of none. It's the kind of thing people say when they're teasing someone who does a bit of everything but doesn't excel at any one thing. But in the age of AI, ducks might just be the species most likely to survive.
Specialists used to win every arena. Deep knowledge was hard to replace. You had to spend five, ten years before you could call yourself an expert. Companies hired for specific skills. Whoever went deeper got paid more. That's how the world worked for as long as I can remember.
Personally, I used to feel bad about myself for being decent at many things but never the best at anything. As a kid I entered art competitions but there was always someone better, so I kept switching styles until I picked up a wide range of techniques. Growing up I played music. Could cover almost every instrument in a band but never mastered a single one, especially when friends around me were better at every piece. Then came work. Design? Sure. Typography and fonts? Can do. Business? Decent enough. Code? Passable.
In the old world where depth was the only metric, that profile is kinda-mid. Nothing to write home about.
But right now, AI is closing the depth gap every single day.
It reads documentation faster than you. Writes boilerplate quicker than you. Debugs patterns it's seen a million times more accurately than you. MIT Technology Review reported that AI already writes 30% of Microsoft's code and over 25% of Google's. Tasks that used to take a junior dev days to grind through, AI now churns out in minutes.
So the question is, what can't AI do yet?
The answer: connecting dots across disciplines.
I've been a designer my whole career. Never seriously worked as a developer (unless you count Pascal, Basic, C, C++, PHP, HTML, and CSS). But once I picked up AI coding tools like Claude Code, Jules, and Google Stitch, I started building Swift apps, shipping landing pages, managing databases. Am I as good as a real dev? Absolutely not. But I can prototype and validate ideas within a single day.
Before, having an idea meant waiting for a dev to help. And waiting some more because they had their own backlog. Now I try it myself first. If it breaks, AI helps me fix it. If it works, I learn why it worked. Then I bring in a dev to refine later. The whole process is ten times faster.
But here's the thing. It works not because AI is smart. It works because someone has to watch over it. I know design well enough to judge whether the output actually works. I know business well enough to tell if the idea has real value. And I know tech broadly enough to communicate with AI precisely.
Not an expert in any single field. But wide enough to connect them all together.
That's a duck.
Now flip the perspective.
People who are deep specialists in one field but have never crossed into others are about to face a serious problem. Because AI can be a specialist too. Cheaper, faster, never sleeps, never calls in sick, never asks for a raise.
But AI can't be a "duck with vision."
It connects dots based on patterns it's seen in data. It doesn't "see" that a design problem can be solved with an engineering approach. Or that a business problem is actually a UX problem. Or that the reason users aren't clicking a button isn't because the button is too small. It's because the copy scares them.
That kind of seeing requires broad experience across multiple fields. The exact thing generalists have been accumulating their entire lives without even realizing it.
Look around and notice.
The people who use AI best aren't the ones who write the longest prompts or know the most tricks. They're the ones who know enough across domains to catch when an output is wrong even outside their main expertise. Who know that a certain approach won't work in a specific context even though they're not specialists in that area. Who know what to ask next because they've seen perspectives from other disciplines before.
A generalist is someone who can QA AI's work across the entire pipeline, from framing the problem to validating the output.
A specialist who can only QA within their own domain will gradually lose ground to AI. Because AI can QA specialized work too.
And it's faster.
The sweet spot
That said, none of this means specialists don't matter anymore. They absolutely do. But the specialists who'll thrive in this era need to be deep in one area and broad across many. The T-shaped kind.
If you're deep but narrow, you're competing with AI head-on. Your price tag will keep getting pushed down because what you do, AI can do too. Not as well yet. But it gets better every month.
If you're wide but shallow across the board, you can use AI but you can't create real value. Because you don't have a deep enough anchor to judge what's good and what's not.
Deep enough to know when AI is wrong, wide enough to know what to ask.
In a world where AI can fill in depth for anyone, the people who can connect the dots will have the edge. Because AI does what it's told, but the person who "knows what to tell it" needs to understand and see the bigger picture first.
So maybe having a short attention span, juggling a dozen hobbies and interests, never quite settling into one lane. Maybe that's not a bug. Maybe it's a feature.
Because a duck fused with AI might just become an eagle.