ā ESSAY
So how am I doing? Honestly, development has become much more enjoyable for me than before. Due to my nature, I'm often not satisfied unless I dig deep into how code works at the foundational level. Before AI emerged, this exploration took way too long, and I often found myself thinking "what am I doing when I should be coding?" and giving up halfway through.
This was particularly challenging when writing React Deep Dive. I couldn't write a single line of the book because I spent so much time digging through code and verifying whether what I wrote was correct. Don't get me wrongāthat time wasn't unenjoyable. I just wished I could have been a bit smarter and understood things faster to accomplish more.
But writing books these days is different. Thanks to AI, my speed of understanding code written by others has improved incomparably from before. In the past, I often struggled to focus on the actual writing because I was busy grasping every detail of the code and verifying its accuracy. Now, with AI's help, I can significantly reduce the time spent on fact-checking and focus more purely on the act of writing itself.
It's like having a smart assistant quickly helping with research. Of course, this smart assistant sometimes brings irrelevant information or misses important context, so blind faith is dangerous. But I can now focus more on the essence of writingāgathering researched materials, distinguishing what's important enough to include in the book, and crafting content that resonates with readers.
This doesn't mean the book quality will dramatically improve though. Writing is still difficult.
This change is equally apparent in my development work. When learning new libraries or frameworks, or when analyzing legacy code, AI serves as a powerful exploration tool. I can now understand code structure and grasp core logic much faster than before, which directly translates to improved productivity.
Beyond just saving time, I now have the bandwidth to attempt deep analyses and various experiments that I wouldn't have dared try before. AI has become an excellent partner that helps me focus more on the essence of what I want to accomplish.
Of course, AI isn't a silver bullet. I somewhat agree with the analogy that "AI is like a junior developer on drugs." AI is incredibly enthusiastic and proactive, never gets tired, and tries to help across multiple domains. But sometimes it reveals biased perspectives from training on biased data, confidently states incorrect information, or suggests code with security vulnerabilities.
Ultimately, verification and responsibility for the output remain entirely mine. Sometimes it confidently spouts such plausible lies that the effort of carefully checking all information still exists. What's important now is developing the discernment to distinguish whether the information this "assistant" brings is actually factual or just hallucinations.
Selecting key insights from AI's information flood, critically examining results instead of blindly trusting them, and steering things in the right direction has become a new core competency for developers. This goes beyond simply generating codeāit requires the ability to effectively question and utilize AI as a powerful tool to enhance problem-solving quality.
I don't think our workload will decrease just because AI helps improve development efficiency. This might actually align with Jevons Paradox. Just as increased efficiency in a particular technology lowers the cost of using that technology, ultimately increasing total consumption of that resource.
As AI automates and optimizes certain aspects of development, we're likely to face more complex and unprecedented requirements, or find ourselves in situations where we need to build more software faster than ever before. However, whether these changes will provide equal opportunities for all developers, or concentrate work among a few skilled developers, or affect corporate workforce management in other ways, requires ongoing attention and discussion.
Continuing this discussion, it's worth remembering that similar concerns and adaptation processes have repeated throughout the history of technological change. When autocomplete first emerged, there were worries that "deep knowledge of libraries would diminish." When IDEs became widespread, voices said "people won't understand the build process." In the 80s, "I can't trust code I didn't write myself" was even a motto.
From this perspective, AI tools like Copilot are ultimately just another incredibly smart tool in this continuum, and their utility will depend on how we use them. Some will strive to learn beyond simple prompting, while others might use AI-generated code in a copy-paste fashion without deep understanding, quickly solving immediate problems but potentially missing long-term opportunities to improve their fundamental problem-solving abilities.
Viewed this way, the AI emerging now might just be another tool to make programming easier and simpler. Just as perfect code editors don't give everyone perfect programming skills, perfect AI won't necessarily create perfect services. But just as we can do more now than in the days of punch card development, we'll be able to realize more possibilities with AI in the future.
Considering this situation where various predictions, expectations, and concerns about AI's future coexist, what's ultimately important is using AI as a tool to elevate our development learning efficiency beyond previous levels and dig deeper into the essence of problem-solving. An era has arrived where I must now accomplish things that were previously impossible due to time or ability constraints.
This doesn't just mean faster skill acquisition, but holistic growth including logical thinking, complex system problem-solving abilities, and creativity. We should delegate simple repetitive tasks to AI while focusing on more creative and essential problem-solving to create new value.
The time gained through AI shouldn't just be turned into leisure, but invested in honing higher-level skills and developing broader perspectives. If we remain at the level of simply creating pages based on React components and hooks as in the past, this means falling behind the pace of the times and might require reconsidering our career as frontend developers.
While all humans fear uncertainty about the future and want to predict and prepare in advance, accurately predicting the pace of technological advancement is nearly impossible. The best choice available to us is to recognize this massive wave of change faster than anyone else and ride on top of it. AI can be a threat, but it can also be an tremendous opportunity. Accepting this fact and actively utilizing it has become more important than ever.
Of course, alongside individual efforts, there's another important variable in the future we'll create: how the organizations we belong toāour companiesāsupport developer growth and redefine roles in the AI era.
Since companies are profit-seeking organizations, we can't ignore the reality that they have strong incentives to focus on short-term efficiency and cost reduction. However, how companies embrace AI, invest in member capability enhancement, and provide growth opportunities may have an even greater impact on developers' futures. Ultimately, how organizations accept this change will determine developers' futures as much as individual efforts.
In this context, some might think, "Now that AI has advanced, aren't frontend developers no longer needed?" or "Our services aren't complexāthey're simple enough that AI could fully replace them." If you genuinely think this way, I'd say the situation is likely one of two possibilities.
First, you're doing business in a truly monopolistic market with no competitorsāa "bloodless victory" situation. While this is a blessed situation, how long it can realistically last is uncertain.
Second, you're creating mediocre, simple services that completely fail to meet users' much higher expectations in the AI era. Here, "simplicity" doesn't mean well-designed conciseness that clearly solves specific user needs, but rather functional and qualitative inadequacy that falls short of market expectations.
If it's the latter, those services won't disappear because AI replaced them, but because they'll naturally be weeded out by the market for failing to satisfy users' elevated standards. Today's users want much more sophisticated and personalized experiences than before, and these expectations continue rising alongside AI technological advancement. The simplicity deemed "sufficiently" replaceable by AI might already be a signal of market rejection.
Therefore, what we frontend developers need now isn't vague fear of AI's emergence or premature expectations that it will "solve everything." Rather, we need a proactive attitude that most actively utilizes AI to learn new technologies and knowledge much faster than before, constantly explores what values are truly needed by our services and users, and nimbly applies them.
This means acknowledging that some existing repetitive coding tasks can be replaced by AI, and instead focusing on high-level capabilities that only human developers can provide: deep user empathy, creative interaction design, and cold judgment of AI output.
AI's limitations are particularly clear in the frontend domain. While AI-generated UI code may look decent visually, it often overlooks accessibility like screen reader support or keyboard navigation, or has inadequate responsive design across various devices. Designing complex component state flows, building design systems where dozens of components work organically together, and optimizing bundle size and rendering performance still require deep domain knowledge and experience.
For example, even if AI generates code to quickly implement specific functionality, we must carefully examine whether this code might cause performance bottlenecks under heavy traffic or accumulate technical debt that's difficult to maintain long-term. Decisions about optimizing data structures or applying more sustainable architecture remain the developer's responsibility.
Additionally, predicting various exceptional situations that can occur in actual operating environmentsāunstable network conditions or unexpected user inputāand potential security threats to design robust defensive logic, as well as thoroughly verifying and optimizing service stability through performance testing and profiling, are tasks only developers can perform.
This means that beyond mere technical proficiency, the importance of soft skills has grown: the ability to define and solve complex problems, communicate and collaborate effectively with colleagues, and find optimal paths among the various possibilities AI presents.
AI isn't technology meant to replace frontend developers, but a powerful tool to help us create better user experiences, solve more complex problems, and ultimately generate greater value. How we use this tool to improve ourselves and create services that satisfy users is entirely in our hands.
Let's ride the wave of change, using AI as our surfboard to gradually move forward. We don't know what waves the future will bring, but this is certainly the most fun and thrilling era for development surfing we've ever had. At least for me, it's more enjoyable than ever.
However, recognizing that this enjoyment may not be equally available to everyone, we must gather our society's collective wisdom to prepare for the shadows that AI's changes may cast.
I hope you'll embrace AI not as a simple threat, but as a collaborative tool and creative partner that can elevate our capabilities to the next level. Designing delicate experiences that exceed user expectations, creatively solving complex interactions, and implementing business requirements in the most visually effective yet performance-efficient ways remain our core competencies.
I encourage you to boldly delegate repetitive tasks to AI and use the time gained to focus on deeper user research, data-driven UI/UX improvements, performance bottleneck analysis, and building highly accessible interfaces.
It's also important to actively share AI use cases and prompt know-how within your team and collaboratively establish review standards for AI-generated code. Code reviews in the AI era will go beyond simple style checking to become processes that verify contexts AI easily missesāservice business logic, user scenarios, and long-term maintainability. Those who share these experiences with colleagues will ultimately go further together.
For those of you nurturing developer dreams in the AI era, these changes might bring both excitement full of new opportunities and some bewilderment. What's certain is that AI is not only redefining your growth path but also how junior developers are viewed and evaluated.
While "how many features you implemented quickly" used to get attention, now "how deeply you understand code and create high-quality output" will make your value shine brighter. What matters is how precisely you ask questions during AI collaboration and how meticulously you verify and improve the results.
Your potential shows in finding the balance between when to wisely seek AI's help and when to dive deep into your own logical thinking during problem-solving. Since AI's information isn't always perfect, experiences in detecting incorrect answers or awkward code and properly correcting them or finding better solutions will be the most solid proof of growth.
AI can be a powerful career accelerator for you. The productivity and autonomy that previously took 3 years to achieve can now be gained in about 1 year with AI's help, enabling rapid growth. Of course, how the gradual reduction of traditional mentoring in this rapid change will affect things long-term, or whether future smaller, more efficient teams might intensify competition, are challenges we all need to contemplate and solve together.
That's why my core message to you remains unchanged: building solid programming fundamentals and concepts is more important than anything else. Add AI's powerful wings to that foundation to absorb complex knowledge faster, try more things, and write your own growth story. Treat AI not as a simple tool that writes code for you, but as a partner that expands your thinking and aids learning. Rather than blindly trusting AI's suggestions, practice reinterpreting and developing them with your own logic. If you maintain an attitude of constant questioning and exploration, the AI era will surely become a sea of greater opportunities for you.
As has been the case in the past, but especially with increased uncertainty due to AI recently, many companies clearly tend to prioritize "senior developer-focused hiring" for immediate project efficiency and short-term utility. Certainly, immediately deploying skilled seniors can bring visible results quickly.
However, this strategy long-term deprives new and mid-level developers of valuable opportunities to learn through diverse practical experience, take on challenges, and grow into the organization's backbone. This leads to disrupted technology transfer within the organization and absence of new perspectives and ideas, hampering innovation capabilities and ultimately becoming a critical risk factor blocking sustainable corporate growth. Therefore, I believe this short-term talent management approach must be reconsidered.
More fundamentally important is breaking away from viewing AI-era developers as "simple repetitive workers who might eventually be replaced by AI" toward recognizing them as "core strategic talent whose roles become more important and expanded due to AI." Cultivating talent at all levels who can effectively utilize AI, control its output, and develop it further is the surest investment in turning AI era uncertainty into opportunity.
Therefore, companies should now focus on redefining roles so developers can effectively collaborate with AI to create new synergies, and make bold investments in building systems where developers at all levels can successfully adapt to this wave of change and grow together. I hope you'll remember that the surest investment for a company's future ultimately begins with belief in and cultivation of "people."