As we enter 2026, discussions about "frontend is going away," "backend will soon be replaced," and "are programmers going to lose their jobs" have hardly stopped.

AI has doubled efficiency, and some people have started to worry that they might be unknowingly left behind by the times.

This time, TinTinLand interviewed three developers from the Web2/traditional IT industry to discuss the real changes brought by AI: improved efficiency, workplace anxiety, and the ongoing professional reconstruction.

• 💬 jethrozz (Java Backend):AI can currently replace the vast majority of frontend work — this is still a relatively conservative estimate. In contrast, AI has had an impact on backend development but has not yet been so disruptive.

• 💬 killzen (Java Backend):Companies have already started to assume that "one person can do more work."

• 💬 zzy (Frontend Developer):What I'm truly anxious about are those leaders who want to reduce costs and increase efficiency in a very short time; they have even developed a kind of blind "AI worship."

Below is the transcript of the interview; let’s listen to their most authentic voices.

Have Frontend and Backend Been Replaced by AI?

👉 AI is accelerating the elimination of mechanized and standardized entry-level positions.

Q1: It's said online every day that "frontend and backend are dead." Which positions do you think are truly in danger?

🔹 zzy (Frontend Developer, 3-5 years of experience)

In actual development, AI still has shortcomings and cannot fully replace human work. For tasks with high originality, the level of implementation is not ideal, and people with professional skills are needed to correct and guide it little by little; however, tasks that are repetitive and mechanical can be completely handled by AI.

🔹 killzen (Java Developer, 5+ years of experience)

Frontend and backend developers will still be in demand, as AI cannot perform humane and specialized functional development.

However, in processes such as development flow, code optimization, and bug fixing, AI assistance is already significantly faster than manual work. In the future, positions involving repetitive labor are highly likely to be replaced by AI.

🔹 jethrozz (Backend Developer, 5+ years of experience)

As a developer who has experience in both frontend and backend technologies, my perspective might be more intuitive. AI can currently replace the vast majority of frontend work (this is still a relatively conservative estimate).

In contrast, while AI has an impact on backend development, it hasn't been so disruptive yet. This is because backend development often relies more on business nature and experiential knowledge, which is harder for AI to help us think through.

Workplace Anxiety and Transformation

👉 Unemployment, rapid change, confusion about career direction... everyone has their own worries in the AI era.

Q2: What is your biggest anxiety regarding AI at this stage?

🔹 zzy (Frontend Developer, 3-5 years of experience)

I don't have much anxiety about AI itself and am even happy that it can greatly lighten my own workload.

What I'm truly anxious about are the leaders who pursue maximum profit and want to realize it in a very short time by reducing costs and increasing efficiency; they even develop a kind of blind "AI worship."

Additionally, for relatively junior developers, they can't participate in the core part of AI technology development while not wanting to continue doing traditional low-value work, finding themselves in an awkward situation of being "neither here nor there."

🔹 killzen (Java Developer, 5+ years of experience)

First, the technological changes brought about by AI are happening too quickly, creating an urgency of feeling that if you don't keep up, you'll fall behind; second, in the face of such drastic changes, I indeed don't know where to transition next.

🔹 jethrozz (Backend Developer, 5+ years of experience)

My biggest anxiety is the risk of unemployment. As AI becomes more powerful, even if it currently isn't effective in some areas, it certainly doesn't mean it won't be in the future. Furthermore, almost every now and then, something disruptive emerges in the industry.

My intuition tells me that any job that relies on computers will eventually be replaced by AI.

The Real Numbers of AI Efficiency Improvement

👉 Behind the technical dividend, "one person doing more work" is quietly becoming the new norm in the industry.

Q3: After introducing AI, how much has your work efficiency really improved? In what specific areas?

🔹 zzy (Frontend Developer, 3-5 years of experience)

About 70%-80% of the code no longer needs to be written by hand. A page that used to take a week can now be put together in a day.

AI's capability in logic implementation is strong, but the UI styling is often not ideal, so developers still have to restore the UI original image or guide the AI to achieve it step by step from single element areas.

🔹 killzen (Java Developer, 5+ years of experience)

Work efficiency has improved by about 30%, specifically in code automation optimization and quick bug fixes.

Moreover, for some technical areas I hadn’t previously tackled, AI can also automatically help me complete the basic work, greatly saving time searching for information, understanding, and filtering.

🔹 jethrozz (Backend Developer, 5+ years of experience)

If we're just talking about writing code and doing business requirements, efficiency has almost improved by 1-2 times.

For example, in my daily workflow, when interfacing new data from a certain XX system, I used to need to sort out documentation and decide which interfaces to connect based on demand, then do the coding.

Now? Documentation sorting, requirement analysis, and coding are all done by AI; we just need to review each stage and debug with a sense of responsibility for the code, and the work is done.

Q4: Is the dividend brought by AI efficiency improvement that you feel more relaxed or more pressured?

🔹 zzy (Frontend Developer, 3-5 years of experience)

For me personally, the dividend lies in being able to quickly expand my knowledge using AI. However, sometimes there may also be an experience of over-relying on AI, leading to disbelieving some existing knowledge.

As for my work state, I'm indeed much more relaxed than before. Our company is taking on many projects; currently, the number of developers is well balanced with the project load, so we haven't started "cost reduction and efficiency increase," but the company won't continue to hire new developers.

Now, after improving efficiency with AI, I can use the time saved to assist other colleagues in participating in different projects.

🔹 killzen (Java Developer, 5+ years of experience)

I haven’t felt too much in the way of relaxing dividends; the most intuitive feeling is that the company has already begun to assume "one person can do more work."

🔹 jethrozz (Backend Developer, 5+ years of experience)

I feel like I’m doing more work. This might be related to my drive; after I finish my work, I continue to push forward or see if other projects need help.

However, within our company and even the industry, the atmosphere for AI coding hasn’t fully formed yet, so I can't really say whether it's become more pressured.

AI’s “Artificial Intelligence Failure” Moments

👉 The last mile of AI still needs to rely on humans to complete.

Q5: What problems has AI struggled with for a long time that you ultimately solved yourself? What tasks would you like AI to handle but still have to do manually?

🔹 zzy (Frontend Developer, 3-5 years of experience)

I previously encountered situations where AI's understanding and implementation were overly complicated; a simple modification of a small portion could achieve the same effect after some review.

Currently, AI's ability to restore images is still insufficient, and sometimes it "overinterprets" or "imagines," so I hope AI can optimize its understanding of images further.

🔹 killzen (Java Developer, 5+ years of experience)

I recently wanted to develop an embedded AI project using ESP32 to connect it to a large language model, enabling simple intelligent interaction between people and physical objects.

However, in practice, programming and testing the esp32 need to be done manually. I hope this area can be replaced by AI in the future.

🔹 jethrozz (Backend Developer, 5+ years of experience)

Right now, I really want AI to handle the verification of my code as well.

My current workflow is to use tools like Superpowers for development, but in the end, I always have to spend time debugging to confirm if the code is correct. CodeX's Computer Use feature can help, but it seems a bit token-intensive.

Additionally, I hope to have another method for logical verification from a data perspective.

Choosing and Avoiding Pitfalls

👉 Choosing the right hardcore practical tools and avoiding concept classes that only sell anxiety is the most basic self-defense strategy for developers.

Q6: What paid AI tools have you used? Which one is the most worthwhile, and which one is the biggest trap?

🔹 zzy (Frontend Developer, 3-5 years of experience)

Currently, I use the paid Cursor for office work. I am learning about AI application development and need to call large model APIs, so I've topped up for some learning token fees.

In comparison, the foreign models (those included in Cursor) are significantly better in terms of coding completeness and more comprehensive in searching the current project files.

🔹 killzen (Java Developer, 5+ years of experience)

I have paid for ChatGPT, Claude Code, and Tongyi Qianwen's large language model. From my personal experience, I find Claude Code to be the most worthwhile.

🔹 jethrozz (Backend Developer, 5+ years of experience)

Both Cursor and CodeX are quite useful. Especially The desktop version of CodeX, which has almost transcended being just a productivity tool for programmers and is also easy for regular users to use.

As for the biggest trap, there's a phrase that says it well: "What is free is the most expensive." For ordinary users, something like OpenClaw might take a lot of effort, but in the end, it may not solve any practical problems.

Q7: There are countless AI courses available now; which types do you dislike the most?

🔹 zzy (Frontend Developer, 3-5 years of experience)

I particularly dislike videos that lure viewers for learning. The code files are already written by the institutions, and the parts participants can write are very simplistic and mechanical; the materials require adding WeChat to access; once obtained, they keep introducing their courses and deliberately create anxiety.

I hope for more genuinely reputable courses that can truly guide students to hands-on practice and build complete projects from scratch.

🔹 killzen (Java Developer, 5+ years of experience)

The AI courses online are indeed overwhelming, and I generally directly search specific technology information practical tutorials from video websites.

The most disliked ones are those with vast concepts but extremely shallow content.

🔹 jethrozz (Backend Developer, 5+ years of experience)

I find it quite annoying when courses use numerous buzzwords during promotion but lack actual valuable content. In this era, people are willing to pay for knowledge, but they dislike being taken advantage of.

Moreover, I also dislike courses without practical application, as they create a sense of helplessness after learning something that has no place to be applied.

If AI Fails, Can You Still Work?

👉 Do you remember the last time you coded by hand?

Q8: If all AI tools become ineffective tomorrow, will your work be paralyzed, slow down, or basically be unaffected?

🔹 zzy (Frontend Developer, 3-5 years of experience)

If AI fails, many stylistic effects I can't complete independently, but the logic part can be implemented slowly, and overall efficiency will definitely decrease.

My current reliance on AI has increased from 50% last year to 80%-90% now. This is mainly due to AI's rapid evolution and more accurate recognition and understanding of natural language and images.

🔹 killzen (Java Developer, 5+ years of experience)

If AI tools become ineffective, my work efficiency will slow down, but it won't affect the final quality of the work.

I only really need AI when dealing with certain specific complex situations. But there's no denying that it indeed saves me a tremendous amount of time.

🔹 jethrozz (Backend Developer, 5+ years of experience)

My work would slow down. I have become used to tossing difficult issues to AI first. However, even without AI, we can revert to using search engines to look up and solve problems.

In fact, currently in my full-time daily workflow, I haven't fully integrated AI yet; it's mostly about "humans driving AI." Sometimes I still need to assist clients with on-site business inspections and communications, in which case there’s absolutely no way to rely on AI, so the dependency for full-time work is not that high.

However, if we're talking about independent small projects that I work on in my spare time, then they are basically driven by AI, and my dependency level there is indeed quite high.

Conclusion

We certainly haven't summarized any "standard methodology for coping with the AI era" from this interview.

Because ultimately, no one really knows the answer.

But "not knowing the answer" has never been a reason not to do anything. What developers are left with is only faster iterations and shorter time windows.

AI skills have become a new dividing line, so let’s put them to use!

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。