AI writes most of the code, what now?
January 03, 2026
This question keeps hitting me again and again. Looking at how fast our profession is changing, this is the truth: AI is writing most of the code.
Technology is moving faster than ever, and these models are just going to get better and better as the months pass. Like it or not, this is our reality now.
After reading countless tweets and articles from top-level engineers, I’ve realized that AI is just taking over the easiest part: syntax and programming. For us software engineers, that was never the hardest part anyway (well, maybe it feels hard when you're a fresher, but you get over it).
I need to focus on solutions and find areas where AI can’t be fully trusted, where human expertise is non-negotiable.
Being a T-shaped developer is now the way to go meaning: have a broad idea of everything, but a very deep understanding of a few core skills.
Here is my personal list of the "hardest parts" that I need to master to stay ahead:
System Designing & Thinking
This isn't going anywhere. AI cannot think strategically about infrastructure, scale, or business value and business constraints. AI can program the modules after I draw the blocks, but when building for scale, knowing every component and how they talk to each other is my job.
Design Thinking (UI/UX)
AI is just a tool, it doesn't have "taste." I need to be the one telling the AI exactly how the UI should look and feel. I am the one who creatively thinks about the layout, the flow of the pages, and making sure the site is actually easy for a human to use
Marketing, Distribution, and Creativity
Now that AI can code anything I want, I should see this as a huge advantage. I can finally build things that are crazy, cool, unique and mine. Creativity always wins. If I build something and don't share it with anyone, then the whole effort was a waste of time. I need to be a builder who knows how to distribute.
The Debugging Mindset
LLMs are not always right. In my personal experience, sometimes they actually "brainwash" you into thinking a certain way is correct when it’s not. So being able to read and understand the code, and beingable to debug given an issue is an invaluable skill. (You are an engineer.)
Security
As AI writes more, the security of the product becomes even more important. I have to be the one responsible for the safety of what we ship.(the product should be secure at every layer, making sure it is should be my job.)
Requirement Gathering
AI can’t talk to stakeholders for you. Even if it tried, it’s hard to trust. The communication gap between what the business wants and what we build has to be near zero.
Using AI as a University
I have tons of knowledge at my fingertips now without looking through 10 textbooks. If I’m creative enough to mix ideas from different domains and pick a problem that needs attention, I can build a unique product and become self-independent.
Technical Writing
AI can help write, but writing with a personal touch, with with beautiful and appealing illustrations and images will always be a level above.
Well, this list can just go on as more oppurtunities in different domains start to rise.
It was always Survival of the fittest and will always be.