Will programmers be replaced in the AI era? The path forward for programmers in the AI era.

Every technological wave brings with it the same anxiety: "Will I be replaced?" "Is this profession doomed?"

Artisans in the steam engine era, skilled workers in the electrification era, and traditional media professionals at the dawn of the internet all asked similar questions.

Today, it's the programmers' turn to ask this question.

But if we shift our perspective from emotion back to reality, we'll find that: **AI's impact on programming isn't about 'eliminating programmers,' but rather re-stratifying them.**

## I. What is AI Truly Changing?

Let's start with the conclusion: **AI is rapidly compressing 'low cognitive density' programming tasks, not 'programming' itself.**

### 1. Areas Already Significantly Impacted

These tasks are being rapidly replaced or marginalized by AI: CRUD-type business code, templated front-end pages, simple scripts, utility functions, documentation-heavy, boilerplate projects, and other coding labor.

The reason is simple: these tasks have **clear rules, short contexts, distinct goals, and low value density**, making them highly suitable for large model generation.

Changes already observed in reality include: a reduction in junior positions, yet increased hiring requirements; a surge in efficiency for individuals or small teams; and 'knowing how to code' no longer being a core competency.

### 2. Areas Temporarily Difficult to Replace

What AI temporarily struggles with is precisely what many have long overlooked: system-level architectural design, complex business modeling, technology selection trade-offs, cross-system and cross-team collaboration, decision-making that takes responsibility for failure, and so on.

AI can provide solutions, but it **cannot bear the consequences for you**, and 'bearing consequences' is precisely the core of an engineer's value.

## II. The True Watershed for Programmers

The original watershed for programmers was from 'people who write code' to 'people responsible for systems.'

The biggest change in the AI era isn't 'how code is written,' but rather **who is thinking and who is executing.**

In the future, programmers will roughly be divided into three layers:

1.  Code Executors: These are the most vulnerable programmers, the ones most easily replaced. This type of programmer has certain characteristics: their primary value lies in 'being able to write code,' shallow understanding of business, reliance on task decomposition, lack of a systemic perspective, etc.

2.  Problem Modelers: Able to transform vague requirements into clear problems, understand business logic and constraints, know 'why it's designed this way,' and use AI as a tool, not a crutch.

3.  System and Direction Designers: Responsible for architecture and long-term evolution, possessing judgment for complex systems, capable of making trade-offs amidst uncertainty, typically not writing the most code, yet deciding the most things.

## III. Practical Advice for Open Source Programmers

Open source programmers often find themselves caught between 'idealism' and 'practical returns.' In the AI era, this contradiction becomes even more pronounced.

Programmers are advised to:

1.  Don't just do 'tech demo-type open source'; shift towards small tools that solve real pain points, infrastructure that can be directly adopted by enterprises, and product-oriented open source that can be embedded into workflows.
2.  Treat open source as 'proof of capability,' not 'emotional consumption.' Capability is more important than the code itself.
3.  Build 'domain tags' early. In the AI era, the value of generalists decreases, while the value of domain experts increases. For example, focus on a vertical domain (e.g., audio/video, AI toolchains, cross-platform, data processing), accumulate long-term understanding of upstream and downstream, so that when someone mentions a certain direction, they think of you.

## IV. Where Are the Practical Paths for Most Programmers?

I must say something unpleasant but true: **Not everyone can become a 'top programmer.'**

Recommended stable and sustainable paths:

1.  **Industry-specific Engineer**: Deeply cultivate a certain industry (finance, healthcare, manufacturing, content), with both technical and business understanding.
2.  **Platform and Tool Builder**: Internal systems, efficiency tools, DevOps, engineering platforms.
3.  **Tech + Product/Management Hybrid**: Technical lead, architect, understands people, understands systems, understands trade-offs.

The real danger isn't 'technology not being new enough,' but rather: **only executing tasks defined by others.**

## V. How to Prepare for This Era, Rather Than Being Swept Along?

1.  Actively use AI, rather than rejecting it. Treat AI as a 'junior engineer,' learn to ask questions, verify, and correct, to improve 'thought output' per unit of time.
2.  Invest in long-term capabilities, not short-term trends. Long-term effective capabilities include: system design ability, abstraction and modeling ability, expression and communication ability, and patience for complex problems.
3.  Accept the fact that 'the proportion of coding will decrease.' Future excellent engineers will: write less code, make more judgments, and bear greater responsibility.

## The Era Won't Eliminate Programmers, Only Those Who 'Stop Evolving'

AI is not the terminator of programmers; it's more like a magnifying glass, amplifying your thinking ability, and also amplifying your voids and dependencies.

What truly determines whether you are left behind by the era isn't whether you can write code, but rather: **Do you understand what problem you are solving, and why you are indispensable?**