South Korean companies are rapidly increasing their use of artificial intelligence (AI), but a new report indicates that the systems to connect this usage to new business and revenue are still insufficient. While the sophistication of AI utilization is advancing, the establishment of commercialization and performance management systems is lagging. The introduction of self-judging and acting AI agents into corporate systems is amplifying the importance of practical business application and operational capabilities.
AWS held a press conference on September 11 at its office in Gangnam, Seoul, to discuss the current state of AI utilization among South Korean companies and the changes that next-generation AI will bring. The report was conducted by global research firm Strand Partners, surveying 1,000 corporate leaders and 1,000 members of the general public in South Korea.
Nick Bonstow, a partner at Strand Partners, stated, "The adoption of AI has become a trend in South Korea. The next challenge is how to spread this rapid adoption into a tangible transformation across the economy."
According to the report, the AI utilization rate among domestic companies rose from 48% last year to 58% this year, marking a 10 percentage point increase. It is estimated that approximately 624,000 companies have newly adopted AI in the past year. The level of AI utilization has also improved, with the proportion of companies reaching the most advanced stage increasing from 11% to 26% during the same period.
While startups (35%) and small to medium-sized enterprises (25%) are leading the advancement of AI utilization, large corporations (9%) are experiencing a relatively slower overall expansion. Only 18% of companies have launched AI-based products and services, indicating a persistent gap between the sophistication of AI utilization and actual commercialization.
Strategic planning and performance measurement systems necessary for commercialization were also identified as challenges. Only 27% of companies have an official AI strategy, and 47% reported that they lack reliable methods to measure the return on investment (ROI) from AI. Although the level of AI utilization has rapidly increased, the systems to expand this into new business and validate performance remain relatively inadequate.
As the spread of next-generation AI approaches, AWS identified a lack of personnel and data as major challenges. Bonstow explained, "The next wave of AI is coming much faster than before," noting that the availability of people and data is a key constraint compared to companies' willingness to adopt.
O Soon-young, a senior solutions architect at AWS, described the structure in which AI agents directly participate in corporate tasks and transactions as the "agent economy." Unlike traditional generative AI, which is primarily used for answering questions or creating documents, agents are utilized to receive goals, plan, and handle actual tasks using necessary tools.
O stated, "AI will not just be a tool; it will become a new economic entity." He explained that AI could expand its role from merely assisting with tasks to making decisions and conducting transactions based on objectives and budgets.
In the agent era, the criteria for assessing the economic viability of AI are expected to change. AWS believes that business viability should be evaluated based on the "cost per completed outcome," which includes inference costs, data and API usage fees, failure and retry costs, as well as human review and security and audit expenses.
Allianz Technology has divided its insurance claim processing into seven specialized agents, reducing the processing time from about 100 days to 20 days. This is cited as an example of how multi-agent-based task automation can lead to actual operational efficiency.
With the expansion of agent utilization, establishing a control system within companies has also been identified as a challenge. AWS emphasized the need to predefine the scope of tasks assigned to agents, their decision-making authority, and accountability for outcomes. O presented the key principles for the agent era as "purpose, authority, and responsibility," stressing that "clarity of direction is more important than the speed of technology."
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.