Efficiency and Optimization Promised by AI:
The Lingering Human Anxiety Behind the System


Why Imperfect Humans Matter More Than Perfect Systems

[Jang Eunsu's World Through Books] On the Imperfection That Makes Us Human View original image

In my neighborhood, there is a bus that operates autonomously using artificial intelligence. It passes by my house about once an hour, and one day, I decided to get on board and observe how it operates. It was not fully self-driving. In potentially risky sections, such as underpasses or where there were obstacles, a human had to drive the bus.


Given the rapid pace of AI development, it seemed likely that someday, there would be no need for human drivers at all. I could easily imagine a day when buses across Seoul would be running without a single human driver. The sight of a person essentially driving themselves and their coworkers out of a job seemed rather ironic. If fully autonomous buses were to become common, would I truly feel at ease boarding one? Or would I be filled with anxiety, worrying about unforeseen accidents? What would become of people who are pushed out of their jobs, forced to give up their skills? These thoughts left me feeling unsettled and uneasy.


There is a concept known as the 'paradox of automation.' The more sophisticated and stable an automated system becomes, the more perilous it is when a human must ultimately solve a problem. British cognitive psychologist Lisanne Bainbridge systematically explained this phenomenon in a 1983 paper. When an automated system is introduced and takes over tasks on its own, the human supervisor loses opportunities for hands-on experience and gradually loses their skills as a result.


Humans are beings who learn and master tasks through coordination between the body and mind. When work that was once performed manually is taken over by machines, personal experience with those tasks diminishes little by little. As a result, people lose their knack for the work, their dexterity fades, and the intricate neural connectivity between brain and body, once operating seamlessly, weakens rapidly, leading to a loss of skill. The mind may remember, but the body cannot follow. This is why those on the front lines often regard their senior managers as incompetent. Unless a person continually works with focus and refines themselves through varied experiences, their abilities deteriorate quickly.


However, when a major crisis such as a natural disaster disrupts an automated system’s ability to function, it is in these critical moments that intensely focused awareness, swift situational judgment, skilled hands and feet, and strong teamwork become necessary. If someone has been away from the practical worksite for a long time, they are likely to panic when caught in such exceptional circumstances. Pilots accustomed to automatic navigation systems and financial workers reliant on algorithmic trading often hesitate and become flustered during emergencies, sometimes leading to major accidents. As automation spreads throughout society with the advent of artificial intelligence and humans and robots increasingly work together in daily life, such incidents will inevitably become more common.


Automation tends to rob humans of their skills and foster incompetence. In fields where automation is implemented, humans are frequently relegated to low-skilled work or reduced to on-call freelancers who are only summoned when needed. The intensity of labor rises, job security declines, and working hours become unpredictable. Such unstable working environments increase the risk of overwork and accidents. This reality drives people into anxiety and even fear—a reflection of our current times.

Concerns are growing that the rapid spread of artificial intelligence and automation technologies may weaken human judgment and skills. Rather than entrusting everything to automated systems, there is a need to leave room for humans to actively intervene and make decisions. The Asia Business Daily Database

Concerns are growing that the rapid spread of artificial intelligence and automation technologies may weaken human judgment and skills. Rather than entrusting everything to automated systems, there is a need to leave room for humans to actively intervene and make decisions. The Asia Business Daily Database

View original image

In his book "On Imperfection That Makes Us Human" (Wilma), British economist Tim Harford argues that automation leads to more frequent mistakes and bigger errors. Automation may enhance efficiency and reduce incidents within an individual work unit, but even if an automated system makes mistakes at a rate only 1/100th that of a human, it can handle tasks a million times faster. In that case, the total number of errors could end up being 10,000 times higher than when humans handle the tasks directly, because machines are not physically constrained by a human’s limits and can work across a much broader scope.


The more a job involves matters of life and death or the risk of large-scale catastrophe—requiring ultra-precise performance—the more dangerous automation becomes. We should not get overexcited just because artificial intelligence achieves a perfect score on a standardized test or excels at playing games. Reality is far more complicated than an exam, and the streets are infinitely more complex than any game. If AI deployed in real contexts encounters something it cannot handle, it could lead to terrible social disasters: planes could crash, power grids might fail, radiation could leak, or financial markets could collapse.


Harford claims that implementing excessively smooth and perfectly optimized systems is misguided. He argues that, for the long term, it is safer to intentionally leave some rough edges or friction—moments when humans must intervene and actively become involved—rather than to have automated systems function entirely without people. This remaining friction allows for resilience even in unavoidable, unexpected situations. Harford quotes social scientist Scott Page: "If everyone thinks the same way, they’ll all get stuck at the same point, and such organizations will fail to move forward."


Thus, just as pilots regularly train in manual flying, we must find ways to preserve human skill—by not entrusting everything to automated systems, but ensuring that people make decisions, perform tasks autonomously, and handle work creatively and spontaneously. Reality is not data, and life cannot be fully ordered. Malfunctions in artificial intelligence and breakdowns in automation systems are inevitable. The people who leap in to resolve these crises must have the highest levels of skill. Yet, those who haven't built up experience solving diverse problems from the earliest stage of their careers struggle to develop the "skilled improvisation" required to respond nimbly and freely in emergencies.


As AI-driven automation technologies become common and states or corporations are increasingly drawn to their apparent appeal, we must remember that reality does not move as neatly as a predetermined script or algorithm. As the original title of Harford's book—"Messy"—implies, the real world is disorganized and full of inconvenient turbulence. It is crucial to value skilled humans who can quickly read situations, judge for themselves, and come up with creative solutions. Only those who have drilled the basics through repetitive practice, frequently encountered problems in uncertain environments, and learned to manage risk through small mistakes and failures, will acquire true skill. Poor organizations focus only on the efficiency and optimization promised by artificial intelligence, whereas good organizations foster skills by forging people in complex and ambiguous situations. Now, more than ever, it is essential to consider what kinds of experiences humans have in organizations and how they spend their time.



Jang Eunsu, publishing and cultural critic


This content was produced with the assistance of AI translation services.

© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.

Today’s Briefing