StartseiteArtikel

"The most outstanding developers are being gradually drained by AI... Having to review 15 PRs every day and painstakingly comb through 20,000 lines of AI-generated code, senior engineers are forced to become "quality inspectors", and some of them simply choose to resign."

CSDN2026-11-18 19:43
Is the job you are doing now still the very one you were passionate about at the very beginning?

AI programming tools are getting increasingly powerful, yet programmers seem to be growing unhappier by the day.

In the past, the biggest headaches for engineers were probably rushing projects, fixing bugs, and staying up late for deployment. Now, AI tools such as Claude Code and GitHub Copilot can already help them generate massive amounts of code, and even take over development tasks that used to take several days to complete.

Theoretically, programmers should have a much easier time.

But the reality is somewhat counterintuitive: some experienced engineers who were once most passionate about writing code are suffering from severe occupational burnout. Some begin to doubt why they chose this career in the first place, some are increasingly losing interest in their current work, and even some choose to resign and leave the software development industry entirely.

Not long ago, tech media outlet LeadDev published a report titled *Engineers grieve a job that no longer exists*, describing this ongoing change spreading among the developer community.

In the report, LeadDev interviewed Annie Vella, an engineering manager and researcher, who said that she has recently met many engineers who are starting to re-evaluate their career choices. At one company, multiple engineers resigned at the same time: some hope that the work of engineers can return to its previous state after they leave, while others simply decide to quit the industry entirely.

And the problem is, they did not leave because they were too busy with work.

Annie Vella said: "Recently, I have clearly noticed that there are more and more posts from engineers on social media. For many people, the focus of work is shifting from writing code to management and verification, such as orchestrating AI agents and reviewing code — which gradually makes them lose their passion for this job."

From writing code to managing AI: Engineers begin to doubt their own value

Many engineers reached out to Annie Vella because an article she published earlier sparked widespread discussion, titled *The Software Engineering Identity Crisis*.

In this article, she pointed out that many people choose to become software engineers not just for a salary, but because they enjoy the process of creating things with their own hands: analyzing problems by themselves, designing solutions, typing code, and finally watching an idea turn into actually running software.

This sense of achievement is exactly the important source of many engineers' professional identity — but AI is changing all of that.

"We are transforming from creators to coordinators, from builders to supervisors. From engineers, we are turning into a role that looks a lot like... a manager."

Annie Vella's research found that 77% of engineers spend less time writing code; nearly half worry that the core skills they once relied on to gain a foothold may be replaced by prompt engineering capabilities in the future.

In the past, the software industry continuously promoted professional division of labor, assigning requirements analysis, system design, testing and other work to different positions, allowing engineers to focus on coding; now, AI is starting to take over more and more coding work, and engineers instead need to spend more time orchestrating AI agents, reviewing generated code, and verifying operation results.

This makes some developers feel a strong sense of loss: they spent years honing their programming skills, but now the most important job has become checking whether AI has written the code correctly.

"I used AI to improve efficiency, but I am losing my passion for work"

This contradiction is not a subjective feeling of a few individuals. Researchers at Oregon State University conducted a study surveying 442 professional developers from 56 organizations to figure out how generative AI will affect programmers' occupational burnout.

The study adopted the Job Demands–Resources Framework to analyze the relationship between work pressure and organizational support after the popularization of AI. The results show that the increased work demands brought by AI, especially the intensification of workload and organizational pressure, will significantly aggravate the occupational burnout of developers.

One developer who participated in the survey put it very bluntly: "With the help of AI, my work speed has indeed increased, and I can complete a large number of tasks in one go, but I am losing my passion for this job."

Moreover, the study also found that there is no significant correlation between occupational burnout and developers' positions, seniority or company size. That is to say, both junior programmers and senior engineers may be affected.

Another *2026 Developer Experience Report* released by Atlassian also presents a similar contradiction:

● Among more than 1,500 software practitioners surveyed, 87% believe that AI can help them deliver work faster, and 80% believe that AI allows them to complete more tasks without extending their working hours.

● But at the same time, 49% of developers spend at least 4 hours a week reviewing AI outputs, modifying prompts, etc., and 75% of developers need to take more responsibility for verifying the quality and security of AI-generated content.

In other words, AI does allow them to complete more work, but it does not make all subsequent work disappear, nor does it necessarily make them enjoy work more.

The most exhausted people are not those who oppose AI, but those who write code conscientiously

In this transformation, the engineers who are most likely to feel tired are not necessarily those who resist AI tools.

On the contrary, many of them were once early adopters of AI tools. They do not cling to the old development methods, but care deeply about code quality, yet they are watching the industry's work standards change right in front of their eyes.

Deedy Das, a former Google and Meta engineer and technical lead who now works as a venture capitalist, once referred to such people as "Craftsmen" in a post. He compared two completely different types of developers:

● One type is what he calls "lazy developers": "They just push AI code up and do not actually code themselves. They do not test manually, and they do not even bother to read the code, leaving everything to automated processes."

● The other type is those "craftsmen" who truly take engineering quality seriously: "The craftsmen are completely worn out, really exhausted."

There are 15 PRs piled up in their review queue, and Slack messages keep popping up. The entire burden of code review falls on these craftsmen. They not only have to review the code, but also take responsibility for it. They are working hard to read the code line by line, make modification suggestions carefully, so that the final delivered product can be better.

However, these people only get a lazy reply from their colleagues saying "this is a great idea", while the modifications submitted with AI still have problems. The craftsmen think to themselves, it's okay, I'll fix it myself, and I even wrote a document to emphasize the importance of code quality.

The next day, new challenges come again: a PR with as many as 20,000 lines of code is waiting for review. Day after day, the craftsmen have more and more review work, and bugs keep entering the production environment, but it seems that no one cares.

Something goes wrong? Just let AI fix it again. The craftsmen are dissatisfied, and eventually they give up: because everything is completely different from the past, the craft they once loved seems to be dead.

As Deedy Das said, the real reason why these craftsmen are tired is not that they are afraid that AI will take away their jobs, but that they have become the last line of quality defense in the team: when no one else cares whether the code has been understood, tested and verified, all the responsibility falls on the few people who do things conscientiously.

The most easily overlooked part is the "invisible labor" brought by AI

Why do managers often find it difficult to notice this problem? Because traditional software engineering metrics are likely to conceal the real situation.

In the past, teams could judge productivity through metrics such as code submission volume, development speed, and deployment frequency; but after AI participates in development, these figures do not necessarily reflect how much labor engineers actually put in — AI can generate a large amount of code in a short time, but this code still needs to be reviewed, debugged and verified by humans. Especially when engineers have to take responsibility for code that they did not write themselves and do not even fully understand, the difficulty of work may even be higher.

And the problem is that this part of work usually does not appear on the performance dashboard.

As a result, an "unfair" situation emerges: people who quickly generate code with AI seem to have high output; senior engineers who carefully review code and fix problems appear to be less efficient because they do not submit enough code.

If the company only focuses on code output without counting the cost of review and verification, it is easy to misjudge the employees who actually undertake a lot of work as inefficient. Therefore, managers need to re-examine the productivity metrics in the AI era, such as counting the time spent reviewing AI-generated PRs, the time spent debugging code written by others or AI, and the energy required to verify the behavior of complex code.

Otherwise, the so-called AI efficiency improvement is very likely to just shift work from one link to another.

Managers, it's probably time to have a talk with your engineers

In Annie Vella's view, the real problem now is not whether software engineering should change with the development of AI, but: is this transformation really for the benefit of engineers, or is it just forcibly pushed forward by the management?

"AI provides engineers with an opportunity to return to holistic problem solving. They can refocus on users, business impact, system design and operation and maintenance work, and become all-around engineers who truly control the entire construction process, rather than just simple code writers."

This is actually a quite optimistic vision, but the problem is that this vision will not come true automatically.

Without proper management and guidance, engineers' work is likely to develop along another path: more and more time will be spent on verification, review and supervision, while people who truly enjoy the creation process will continue to leave.

For managers, this is no longer the familiar occupational burnout they used to see. It is no longer just the fatigue caused by too much overtime, tight project deadlines, or frequent on-duty shifts, but the very nature of work is changing, at a speed even faster than people's ability to adapt.

Many engineers who are experiencing occupational burnout have long been going through a hard time. It is not that change itself is a bad thing, but that people rarely care about the direction this change is heading, and what they are experiencing right now.

For engineering managers, the first step is very simple, yet very difficult to do —

Is the work you are doing now still the job you loved in the first place?

Original link: https://leaddev.com/career-development/engineers-grieve-a-job-that-no-longer-exists

This article is from the WeChat official account "CSDN", compiled by Zheng Liyuan, and released with authorization from 36Kr.