Frontend Jobs Are Disappearing Everywhere: AI Collapses the Front/Back Split Into 'Full-Dry' Roles

---

theme: smartblue

A while ago I wrote an article whose core point was: “The frontend job is gone now.”

But there are always people who like to argue, saying I’m only talking about big companies, and that most people work for small and medium-sized enterprises (SMEs), which have only trimmed some staff, but the role still exists.

Reality

To be honest, I find this line of argument somewhat “sad.”

Image from the internet, contact to delete if infringement

What can I say? I agree with the first half, but definitely not the second half.

It’s true that most people are employed by SMEs, and I don’t think that’s wrong. Of course, the statistics also support this.

According to the Fourth National Economic Census, small and micro enterprise legal entities account for 99.8% of all enterprises, and they absorb 79.4% of all enterprise employment.

As of September 2025, the number of registered SMEs nationwide reached 63.487 million, contributing more than 80% of urban labor employment.

Although big companies recruit in large numbers, the total number of positions is far smaller than that of SMEs.

But that doesn’t mean the “frontend job” still exists!

Let’s first talk about the big companies. Among the well-known ones in China — ByteDance, Alibaba, JD.com, Tencent, Meituan — some have already leaked news, some haven’t.

But the removal of frontend positions and frontend departments is already the general trend.

You say SMEs will follow the pace of big companies, but would big companies follow the pace of SMEs?

I’m not saying the frontend work is gone; the job title is gone. The work still exists.

Some frontend bugs still need someone to debug, some frontend pages still need someone to modify, and some frontend business logic still needs someone to code.

AI has increased the speed of software development by more than an order of magnitude. If before it was riding a bicycle to Tibet, now it’s riding a rocket!

Impact

Many people also say: AI-written code is far less robust than human-written code.

Image from the internet, contact to delete if infringement

In some scenarios I agree with that, but in the vast majority of scenarios I don’t think it holds!

Right now, most people’s code doesn’t even do basic type checking, while AI-written code almost always includes type checks on all parameters.

If you don’t believe it, ask it to implement an array sorting method. You might write this:

function arraySort(arr) {
    return arr.sort((a, b) => a - b);
}

And the AI’s sorting method would likely be:

function sortArray(arr) {
  if (!Array.isArray(arr)) {
    throw new TypeError('The argument must be an array');
  }
  return [...arr].sort((a, b) => a - b);
}

In the past, before adopting TypeScript, I spent countless times debugging similar errors. TypeScript made things a bit better, but there were still some “big brothers” in the team who would write any.

Industry

AI is strong, and I think there’s nothing to argue about that.

AI is developing fast, and this is obvious to everyone.

Image from the internet, contact to delete if infringement

As for AI’s impact on the industry, there’s no need to mention it. The two most direct effects on the entire software industry right now are: delivery time and layoffs.

How long did it used to take to develop an OA system? Even without too many customized features, it took about a month, right?

Now? Doing custom work takes about two weeks, and that includes time for requirements gathering.

The good side is “customer satisfaction,” and the bad side is “project out of control.”

Once the delivery schedule gets too tight, programmers essentially stop reading code. They mechanically type in requirements and wait for AI to output code.

The whole process becomes:

Requirements intake, translate requirements to AI, AI outputs code, AI tests, go live.

In this process, the programmer barely touches the code, and nobody knows what AI actually implements in its internal logic. That is, the code is all out in the open, but the entire feature’s implementation becomes a “black box.”

• Because the customer doesn’t read code — no need, they want the functionality;
• Because the project manager doesn’t read code — they care about the deadline;
• Because the programmer doesn’t read code — when the schedule is tight, who has time to study how it’s implemented? Just implement it;
• Because the tester doesn’t read code — testers have long been laid off.

Everything is a “black box,” and that’s how it goes live. Problems are just a matter of time; the only reason they haven’t appeared yet is that it’s not the right moment, but the hidden dangers are already planted.

I’m not saying human-written code has no hidden dangers, but with human-written code, the person can roughly know why a problem occurs.

With AI-written code, nobody knows. In the end, it comes down to: as long as either the program or the person can still run, that’s fine!

Answer

Back to the original question: Does the frontend job still exist?

With AI’s help, writing code has become an extremely simple thing. Instead, programmers spend most of their time on “requirements,” telling AI what to build.

Image from the internet, contact to delete if infringement

So, can a frontend programmer describe backend requirements?

If they can, then the backend engineer is not really necessary in that project.

Then, can a backend programmer describe frontend requirements?

If the answer is also “yes,” then the frontend engineer is also not necessary in that project.

That’s why the job title “full-stack development” from the era before frontend/backend separation has been revived. Because AI has erased the technical barrier between frontend and backend; we are essentially translators of AI requirements, not writers of code.

Why did those big companies propose removing the frontend job earlier? Perhaps because you are a frontend developer, you see more news about frontend positions being cancelled.

Actually, backend positions and testing positions are also being “agglomerated” in large numbers. I guess there are ultimately two paths:

• Programmers who specialize deeply in a specific area, such as frontend million-data Canvas rendering, WebGL, complex industrial scene simulation.
• Full-stack programmers (or “full-dry” programmers, as I call them)

Conclusion

The times are moving forward. Whether you admit it or not, whether you want it or not, it will not stop for you for a single second, not even a brief second.

We might be the king of our own world, the protagonist of our own world, the pillar (the treasure) of our family. But to the entire industry, we are just a grain of sand, or even dust.

Image from the internet, contact to delete if infringement

All we can do is take care of ourselves, and while others are being eliminated, do our best to keep going, so that we can delay our own elimination for a little longer.

Does it feel like I’m saying “comparing who is worse” or “involution”?

Exactly!

I am Li Jianyi, a person who hopes to understand himself clearly. Welcome to follow my WeChat public account of the same name, “李剑一”.