At 11 p.m. in a law office in Jongno, the sound of typing fills the air. First-year attorney Kim James, 37, substitutes a sandwich for dinner as he continues working late into the night.
"These days, if you can't get work done, it's not that you get scolded; you just don't get assignments. If that happens repeatedly, you could lose your job," he said. Balancing corporate advisory and litigation, late nights have become routine for him. "Even at 1 a.m., the atmosphere in the office is intense," he explained. He actively uses AI for issue organization and case law research.
Kim noted, "With clients using AI, if lawyers don't, they can't keep up." AI has deeply penetrated South Korea's professional sectors, threatening entry-level jobs. He observed, "As research tasks are increasingly replaced by AI, law firms are hiring fewer new associates."
According to the Bank of Korea, from July 2022 to July 2025, about 211,000 jobs for young people were lost, with 98% concentrated in sectors highly exposed to AI. By February 2026, employment in professional, scientific, and technical services saw a decline of approximately 105,000, marking the largest drop since records began.
Kim considers himself fortunate to have found a job. "For my juniors, it's more about 'escaping' than finding employment. They attend bar association training while repeatedly submitting applications," he said.
He feels threatened by the rapidly evolving AI landscape, describing the change as "an opportunity for a few but a crisis for many." He emphasized that the gap between lawyers will widen based on their ability to utilize AI, stating, "Ultimately, what matters is the questions you ask AI, which requires fact-finding and communication skills with clients."
"Results remain, but learning disappears"
"The way new professionals grow into experts is changing," said Professor Kwon Hyuk-koo from Nanyang Technological University in Singapore.
In the past, newcomers learned judgment criteria through repetitive research and drafting, but AI is replacing that process, creating a structure where "only results remain, and learning disappears." Kwon described the repetitive tasks assigned to junior staff as a "training ground" for learning the industry. He warned that companies treating AI merely as a cost-cutting tool might achieve short-term efficiency but weaken their talent pipeline in the long run.
As AI rapidly spreads, the roles of humans and AI are becoming increasingly distinct.
Architect Kwon Mo, 37, who leads an AI department, noted, "In recent years, the way we work in design has changed significantly due to increased AI utilization." He remarked that initial tasks like image generation and design alternatives are now much faster, greatly enhancing efficiency. However, he added, "Even with tools like Nano Banana or Midjourney, refining prompts and translating requirements into actual designs still requires human judgment."
Joshua Yoo, a lawyer at a Seoul law firm, observed that as contract review and research tasks are replaced by AI, the process for new hires to learn "ways of thinking" has vanished.
Kwon believes these changes are reshaping the hiring market. "The industry has cut entry-level hiring by about half, while demand for professionals with 5 to 10 years of experience is rising. Now, the ability to use AI and digital tools, along with collaboration and planning skills, are crucial hiring criteria," he explained.
Challenges in Education and the Rise of 'Verification Skills'
Kwon pointed out that if AI takes over tasks previously handled by junior staff, the key challenge will be how young professionals develop their judgment skills. He emphasized that universities must teach not only AI utilization but also critical verification and the integration of human judgment. Additionally, he noted that adult learning participation rates are lower than the OECD average, highlighting the need to strengthen lifelong learning and retraining systems.
Professor Lyse Langlois from Laval University in Canada agreed that the automation of entry-level tasks has weakened the traditional pathway for skill development. She warned that failing to critically review AI results could lead to a state of "autopilot," where individuals follow results without independent judgment. Erik Cambria, also from Nanyang Technological University, predicted that the ability to understand concepts and relationships will become more important than simple task-oriented learning, emphasizing that the capacity to critically verify AI results based on context and human understanding will be crucial.
In contrast, the healthcare sector is experiencing a "soft landing" with AI, as it reduces repetitive tasks and allows medical staff to focus on judgment. In March 2026, South Korea's employment trends indicated that the health and social welfare services sector added 294,000 jobs, the largest increase among industries, while professional, scientific, and technical services saw a decrease of 61,000 jobs during the same period.
The relatively smooth integration of AI in healthcare is attributed to its role as an augmentation tool that assists rather than replaces human capabilities. As demand for healthcare rises due to aging, employment remains stable or increases, with AI helping to streamline repetitive tasks, allowing medical professionals to concentrate on patient care.
Benjamin Lee, a 36-year-old resident at a university hospital in the capital region, stated, "AI has definitely made my job easier." He noted that fields like radiology and emergency medicine have seen significant changes, with AI quickly assisting in analysis and prioritization, automating tasks like patient scheduling and record-keeping, enabling doctors to focus more on clinical judgment.
Proposed Solution: 'AI Apprenticeship Model'
Experts suggest that universities should enhance the 'AI apprenticeship' model in education. Kwon stated, "Instead of simply submitting AI-generated results, we need to teach students what prompts they used, what they verified, and why they accepted or rejected certain outputs."
Langlois emphasized that whether innovation is maximized or if AI serves as a supportive tool for everyone ultimately depends on human choices. Cambria added that education should be redesigned to go beyond technical skills, fostering a critical understanding of the relationship between humans and AI to protect oneself.
As the clock nears 2 a.m., a design office in Gangnam remains lit. Monitors display design materials and AI tools, while coffee cups pile up on desks. Amid the rapidly generated outputs, the responsibility for choices still rests with human judgment.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.