UX in the AI Era and the Role of Designers: Insights from UX Korea 2026 Fall

Sola from the Business Strategy Division at QUANTUM C&S attended UX Korea 2026 Fall, which took place on September 2.

This event provided an opportunity to explore how user experience and the role of designers are evolving in the AI era through a variety of topics, ranging from AI agent experience design to UX knowledge, AI-based UX research, ontology, physical AI, and AX design.

Although the topics and approaches of the various sessions differed, one common question recurred throughout the day.

As AI creates and performs more tasks, what should humans focus on?

From creating to judging

Generative AI is already rapidly producing a variety of outputs, including screens, documents, and code.

The LG Electronics session addressed how, amid these changes, the importance of evaluating, curating, and refining AI-generated results is growing—outweighing the time spent on hands-on creation.

In this process, the role of the designer is also gradually changing.

Whereas in the past, the ability to personally produce high-quality results was crucial, in the future,

the ability to define what kind of experience to create,
establish criteria for what to prioritize,
and determine which AI-generated outputs to select

may become even more important.

A similar perspective was presented in Samsung SDS’s Agent Experience Design session. The message was that a good AI agent goes beyond simply using an outstanding model; it requires designing the entire experience—including the agent’s role, the scope of human intervention, guardrails, and grounding.

AI Also Needs an Organization’s ‘Criteria for Judgment’

A particularly interesting part of this event was the discussion on how to create knowledge that AI can utilize.

Connecting a design system to AI allows it to use predefined components and styles. However, even when using the same materials, deciding what information to display first, what to omit, and which patterns to choose based on the situation is a separate challenge.

To address this, LG Electronics presented an approach to structuring the decision criteria—which had previously relied on individual designers’ experience and intuition—into explicit UX knowledge that AI can understand.

For example, instead of simply

“Use a step UI when there are many steps”

,

“In situations where there are many input fields and a high risk of user drop-off, use a step-by-step UI to reduce cognitive load,”

—a method that documents both the situation and the rationale behind the decision.

This approach extended beyond design to the company’s overall operations.

AI That Understands “How We Work,” Beyond Corporate Data

During the Enhans session, it was explained that for enterprise AI to perform actual tasks, simply connecting data is not enough.

This is because each company uses different terminology, and even with the same data, judgment criteria and business rules vary.

The concept introduced to address this is Ontology.

An ontology is a structure that connects not only the entities and relationships within the data but also the criteria by which an organization interprets situations and makes decisions, enabling AI to understand them. The presentation likened this to a “map explaining the company’s business world.”

What was particularly striking was the point that, in building this knowledge, the process of uncovering the tacit knowledge of business practitioners is just as important as the technology itself.

“Which data should be examined first?” “At what
level is a situation deemed risky?” “When do exceptional
situations occur?” “Who makes the final
decision?”

Through these questions, the judgment criteria that existed within human experience are transformed into an explicit structure.

Ultimately, we confirmed that AI transformation goes beyond merely adopting technology; it begins with the process of understanding how an organization actually works and makes decisions.

A New Approach to User Research Using AI

NAVER introduced a case study of LLM-based virtual persona UX research called EchoX.

UX research is a crucial process for gaining a deep understanding of real users, but it requires significant time and resources for participant recruitment, interviews, and qualitative analysis.

EchoX generates a virtual persona with desired characteristics and simulates that persona navigating and interacting within a real web environment. Researchers can then analyze the behavior logs and results or even ask the persona questions directly.

This approach is significant not as an attempt to completely replace real users with AI, but as a new tool that complements early hypothesis exploration, idea validation, and UX research.

A similar study on AI replicating users was also presented at the Yonsei University session, but it highlighted a limitation: even though AI can generate plausible responses, it cannot perfectly substitute for the emotions or pain points of real users.

This reaffirmed that the more important question is not how much AI we use, but rather in which areas we should trust AI and where we should reserve judgment for real people.

The Challenge After Generation: “Validation”

As the speed at which AI generates results increases, verification becomes increasingly important.

During the AXis session, it was explained that simply having AI produce good results is not enough to apply them to actual work; clear criteria are needed to determine which results should proceed to the next stage.

Rather than reviewing all results at once and asking, “Is this okay?”, the workflow should be broken down into small steps,

defining what must be met at each stage → having the AI verify it → and having a human make the final judgment on high-risk results

.

In other words, even in AI-driven automation, the goal is not to eliminate human verification; rather, it is becoming increasingly important to more clearly define the points where human judgment is absolutely necessary.

Even in the AI era, it is people who design experiences

The changes observed at this year’s UX Korea 2026 went beyond the mere emergence of new AI design tools.

As AI’s ability to produce outputs rapidly advances, the very scope of UX and design is shifting

expanding from screens → agents → workflows → to an organization’s knowledge and decision-making systems

.

At the same time, a common theme emphasized across multiple sessions was the role of humans.

Providing context and criteria for AI to act upon;
deciding how much to entrust to AI;
evaluating and verifying the results produced by AI;
and imagining new experiences that do not yet exist.

As AI takes on more tasks, the role of defining problems and making judgments is likely to become even more important.

QUANTUM C&S will also continue to distinguish between the areas where AI and humans excel, and to explore how to design experiences and systems that allow AI to function effectively in support of the actual work of people and organizations.