How AI Big Models are Changing the Way Programming is Done

To be honest, when I first saw how well the Gemini 3 Pro performed at programming tasks, I was blown away. As someone who works a lot with code, I never thought programming could become so "spontaneous". In the past, writing code was like carving with a chisel, but now it's like shaping it with your mind - all you need to do is describe what you want, and the AI will help you turn it into reality. This change is not just an increase in efficiency, but a revolutionary shift in the entire programming paradigm.

From precise instructions to fuzzy requirements

I remember when I first started learning programming, my teacher always emphasized that the syntax had to be accurate down to every semicolon. How about now? You can tell an AI in natural language, "Write me a script that automatically organizes my photos, sorting them by when and where they were taken, and also recognizing blurry photos." And then it literally generates runnable code! This shift has lowered the bar for programming dramatically, but it has also created new challenges - now we need to learn how to accurately describe the problem more than we need to rote memorize syntax rules.

One particularly interesting example: a designer friend, who knows absolutely nothing about programming, used Gemini 3 Pro to make an interactive portfolio website. She just described the desired animation effects and page structure, and the model generated the complete code. This would have been impossible before!

Code Quality Redefined

In traditional programming, we spend a lot of time debugging and optimizing. Now AI can not only generate code, but also automatically optimize performance, add comments, and even refactor code. But then again, this complicates the question of what makes good code. After all, when AI can easily write code that meets specifications, what is the unique value of a programmer?

I've recently noticed an interesting phenomenon: teams using AI-assisted programming are shifting the focus of code review from syntax checking to architectural design and business logic. This may be the way of the future - humans are responsible for grasping the overall architecture and creative solutions, and AI is responsible for implementing the details.

Disruptive Changes in Programming Education

Programming instruction is still teaching basic concepts like loops and conditional judgment, but what about five years from now? Maybe the focus will become how to collaborate with AI, how to design algorithmic ideas, and how to evaluate AI-generated code. This got me thinking, what new skills will the programmers of the future need to have?

Here's a telling statistic: after using AI programming tools, junior developers are on average 3x more productive, but senior developers are only 1.5x more productive. This shows that AI is leveling the skills gap, making creativity and problem-solving skills more important than coding skills.

To be honest, I'm both excited and a little apprehensive about watching these changes. Programming is changing from a technical job to a creative one, which is perhaps the inevitable development of technology. Just as painting didn't disappear after the invention of the camera, it just shifted from recording reality to expressing art. The future of programming may be the same.