What I Observed After a Month Working with AI
After about a month of daily collaboration with AI, I've developed a strong concern. Namely, "how should corporate talent investment strategies change in an era where AI can completely substitute junior-level work?"
During this month, I spent several thousand dollars on Claude. From a management perspective, it's nearly impossible to hire such a competent worker at this price. That's precisely why, when compared to hiring humans, I've rationalized that this is absolutely worth the cost.
Witnessing AI capabilities firsthand has made me strongly realize that the economic value of junior-level talent is rapidly declining. This isn't merely a technical issue, but a fundamental shift affecting business management and talent development.
Redefining Talent Value in the AI Era
The Distance Between Juniors and Problems
AI will first replace junior-level work. This is because juniors are positioned "furthest away from the problem."
The core of business always lies in "solving human problems." Seniors directly engage with customers, understand on-site challenges, and organize these into requirements. Meanwhile, juniors handle "systematically executing defined tasks based on requirements that seniors have organized from issues brought in by sales."
AI's greatest strength is its "ability to efficiently execute defined tasks." It works cheaper than humans, without complaints, promotion negotiations, interpersonal conflicts, or criminal behavior. It operates 24 hours a day, can easily scale to the workload of 100 people, and possesses knowledge far superior to the average person.
A Lucky Generation for Seniors
Those of us with substantial professional experience are quite fortunate at this juncture. As seniors, we're already practiced in delegating work to juniors, so our roles remain largely unchanged when this work shifts to AI. We already understand how to create value.
Fortunately, having invested considerable time and been allowed to make many mistakes, we can lay tracks for AI to avoid these same mistakes, creating a surprisingly effective division of labor with AI.
AI can write functioning code, but it cannot determine specifications. Specifications arise from business challenges, where human intervention remains highly valuable. As long as the source of problems remains human, humans will always be needed there. How close you are to the problem becomes your value. From this perspective, juniors are positioned furthest from the problem.
Mastering AI Means Growing Yourself
Through collaborating with AI, I've deeply considered what it means to "master AI." Those celebrating improved AI capabilities should verify whether their own abilities are also improving through AI. "Mastering AI" means nothing less than maximizing your own capabilities using AI while drawing out AI's maximum potential.
In my experience, you can't simply delegate to AI and sleep soundly. Rather, to bring out AI's power, my brain has been operating at full capacity lately, making me daily aware of my own limitations and growth potential. I thought I was putting AI to work, but ironically, AI has been putting me to work—yet I'm grateful because I feel my own growth.
Though AI possesses advanced computer science and systems engineering knowledge, if the user's level is low, output remains limited to "Hello world." High-quality, sophisticated questions elicit correspondingly advanced responses, and vice versa. In essence, human capability significantly impacts AI capability.
How AI is Transforming Work
AI Begins by Complementing Existing Industries
When considering talent strategies that leverage AI, it's important to recognize that AI initially replaces only portions of existing jobs. My experience collaborating with AI suggests that AI has remarkable compatibility with tasks previously assigned to juniors.
Services used by more than 100 people—potentially even millions—should continue receiving proper investment and human oversight, as development costs remain justified. Everyone can share the payment burden. This is an economy of scale, similar to how larger apartment buildings can afford shared spaces and grand entrances because the per-person cost remains manageable.
The Potential for Drastically Reduced Implementation Costs
With Vibe Coding (a collaborative AI coding style based on conversational flow), while maintainability might be lower from certain perspectives, implementation costs drop so dramatically that the range of applicable work expands exponentially. Manual tasks that once took 8 hours might be implemented with AI in 30 minutes, potentially completing the entire job within an hour. I've personally experienced these benefits numerous times with data preprocessing.
When implementation requires minimal time, the applicable scope expands dramatically. Automation for just a few people might become cost-effective. You could even write code solely for yourself.
Rebuilding Strategies for the AI Era
How Should Companies Reshape Their Talent Investment Strategy?
In this context, how should companies approach talent investment? Here's my perspective:
-
Evaluate talent by proximity to problems: Prioritize investment in people who create value from positions close to customer problems and business challenges
-
Emphasize AI collaboration skills: Value the ability to effectively collaborate with AI over mere task execution capabilities
-
Redesign junior development models: Build new development approaches that presuppose AI collaboration rather than traditional OJT models
-
Maintain a medium to long-term perspective: Consider organizational culture and tacit knowledge transmission alongside short-term efficiency
Crucially, what AI replaces is "tasks," not "problem-solving abilities." What businesses truly need are people who understand fundamental business challenges and can solve them by leveraging various resources, including AI.
The Crisis for Junior Talent and Career Strategy
Now that AI operates with performance completely surpassing junior-level work, finding justification for hiring junior-level personnel has become increasingly difficult. For those seeking employment without experience, to get hired without experience, you must demonstrate ability to collaborate with AI rather than position yourself as replaceable by AI.
While I generally avoid definitive expressions like "essential," I want to emphasize this point strongly given the crisis, so I'll state it definitively: this is essential. The traditional mindset of expecting companies to train you won't work when AI presents a better alternative. Even without experience, you must practice demonstrating value through AI collaboration.
The Essence of Career Building and Sustainable Growth
Developing Talent in the AI Era: Kikagaku's Perspective
This isn't exclusively about AI collaboration. The principle applies to all professions. Attending some school once a week for a few hours will never make you first-rate. The world isn't so forgiving that such superficial abilities would secure you a new job.
Certainly, until now, AI and data science fields received attention due to labor shortages, creating the impression that developing minimal skills could secure higher-paying positions. But that was never really the case. Working in job placement support, I've observed patterns among those who succeed or fail after studying data science. Those who add data science (secondary) to their existing strengths/domains (primary) tend to succeed in transitioning within their primary field. They aren't making radical career changes.
A common misunderstanding occurs when people believe they can transition into an entirely new field as a data scientist. Such individuals often lack a primary strength. While they might want to develop data science as their primary strength, having spent two years in graduate school at Kyoto University, let me be clear: it's not that easy. Before entering graduate school, I studied for over 10 years, survived the competitive Kyoto University entrance exams to enter my desired laboratory, and then experienced two years of hellish living. Unable to work part-time jobs, I spent 24 hours a day thinking about research. After this roundabout journey of constant thinking and endless discussions with world-class professors and students, I've now spent another 10 years in my professional career.
At Kikagaku, the company I founded, we've consistently communicated this reality. Witnessing individual-focused schools making money by funneling graduates into System Engineering Services (SES) through job placement saddened me, so we've been transparent about prerequisites from the beginning. Of course, people only see and hear what they want to see and hear, so problems never disappear entirely, but I believe we've operated as conscientiously as possible.
The 3000-Hour Rule: The Importance of Fundamental Learning
I believe developing these skills requires "3000 hours." Working 10 hours daily for 25 days monthly equals 250 hours. Continue this for 12 months, and you reach 3000 hours. In research circles, I've often heard that "3000 hours of focused work makes you first-rate in that problem domain." While I can't verify this claim, dedicating a year to intense focus should certainly bring you close to mastery among peers working on the same problems.
This 3000-hour commitment isn't easy. The first month might feel like smooth progress on the growth curve (the famous Dunning-Kruger effect where confidence exceeds knowledge), but the subsequent plateau is overwhelming. Persistence through this period is what brings you closer to mastery. Most of this time will be filled with pain. Whether you can continue determines your outcome.
If you want to work professionally, start by reforming your mindset to accept at least one year of personal training without expecting short-term results. If you can immerse yourself in something for 3000 hours, it should become your powerful ally. If you can't dedicate 3000 hours, don't force yourself—other paths exist.
Personal Career Strategy: Growing Alongside AI
Using AI as both ultimate teacher and ultimate partner, creating prototypes and bouncing ideas off it accumulates valuable insights. It's about choosing between doing what fundamentally matters versus resorting to superficial job-hunting techniques that diverge from contemporary realities.
Conclusion: Face Change Without Fear
My month collaborating with AI has strongly impressed upon me that AI evolution cannot be stopped. And this change is already significantly impacting junior-level work.
For companies, reconsidering the economic rationality of developing junior talent and creating new talent strategies centered on problem-solving abilities and AI collaboration skills is necessary. For individuals, this requires investment and determination to stand alongside AI rather than be replaced by it.
As an educator, I view this change positively. While traditional junior development models may collapse, collaboration with AI should enable value creation at higher dimensions. What matters is facing change without fear.
Collaborating with AI also expands human potential. I look forward to continuing discussions with all of you about talent development and personal growth in this new era.
