[THE VIEW]The Value of Deciding to Stop AI

Who Decides What to Power Down?and by What Standard?Defines Competitiveness
Not What to Automate,
But What Never to Automate

Just a few years ago, it was novel enough for a company to simply claim it had adopted artificial intelligence (AI). If the word "AI" appeared in a press release headline, it drew attention; if AI transformation was mentioned at an investor briefing, questions would decrease. But things are different now. Saying a company uses AI no longer serves as an explanation. It is closer to a default, and what matters now is not whether AI has been adopted, but what has changed after its adoption.

Recently, an interesting shift has been observed among some American companies. Rather than adding more AI features, there is a trend toward reducing or integrating them. For example, instead of placing chatbots on every service screen, companies are now limiting them to only core work areas. On the surface, this may appear to be a step backward, but internally, it is described as a phase of consolidation. It signifies a transition from a period of experimentation to a period of selection.

In this context, the term "AI washing" frequently appears. Specifically, this refers to the phenomenon of attaching phrases like "AI-based" or "AI-driven" regardless of actual results. The problem is that this is not just marketing exaggeration. AI washing clouds judgment within organizations. Once the perception takes hold that "we are already using AI," more important questions disappear-questions such as whether this technology is truly necessary, whether it is suitable for the current work, and who will take responsibility.

An executive at an American company put it this way: "We didn't start using AI better; we started using it less." This means they prioritized identifying what not to do with AI, rather than expanding what AI could do. Many of the features that had been piloted across areas like customer service, internal document summarization, and decision support were removed. Usage rates were low, maintenance costs were higher than expected, and most importantly, employees did not trust the results.

The key here is not the performance of the technology, but the strategy. AI is a general-purpose technology, but strategy is a matter of choice. AI does not become more powerful the more widely it is used; it is a tool that aids organizational judgment only when applied appropriately. We have reached a point where the decision not to use AI in certain areas requires more explanation than the fact that AI is being used.

Instead of turning on all functions, choosing to leave AI only at points where responsibility can be taken has become important. The moment of deciding where to stop has become more crucial than using AI itself. Gemini generated image

The common trait among companies that survive after the era of AI washing is clear. It is not about how well they explain their use of AI, but whether they can articulate why certain areas were left untouched by AI. Why did they keep people in this particular work? Why was this decision not automated? Why was this experiment halted? Only organizations that can answer these questions can claim to have control over the technology.

Korean companies stand at the same crossroads. Whether or not AI has been adopted is no longer a differentiator. What matters more is whether meetings have decreased after adoption, whether decision-making speed has changed, and whether responsibility has become clearer. How many times the word "AI" appears in a report is meaningless. If the way the organization operates has not changed, it is closer to ornamentation than true adoption.

Competitiveness in the AI era does not come from having more features. Rather, it comes from the ability to determine what to remove and where to stop. AI will not pass by as a fleeting trend, and whether to adopt it is no longer a viable option. Over time, AI will become cheaper, more widespread, and more deeply embedded within organizations.

As this happens, the basis for competition will move further away from the technology itself. In an environment where every company can access similar levels of AI, what sets them apart is not what they choose to automate, but the consistent standards for what they choose not to automate, even to the end.

After the era of AI washing, the companies that remain will not be those that adopted AI in the flashiest way, but those that calmly consolidated, were honest enough to halt projects when necessary, and steadfastly maintained their own principles for coexisting with AI.

Son Yoonseok, Professor at the University of Notre Dame (USA)