7 in 10 large firms say basic AI skills enough for new hires

Seven in 10 large South Korean companies say they expect new hires to have only basic AI skills — such as using AI for search and summarization — while more than half of job seekers believe they need advanced capabilities, including verifying AI outputs and redesigning workflows, to land a job.

The Korea Economic Forum (KEF) released the findings Tuesday from a survey of 200 companies among the country's top 1,000 firms by revenue and 500 job seekers — university students in their third year or above, or recent graduates from four-year institutions.

According to the survey, 77 percent of job seekers said AI proficiency is important for employment, compared with only 39 percent of companies. However, 83 percent of the companies surveyed expected the level of AI competency required of new hires to rise over the next three years, pointing to growing expectations ahead.

The gap between job seekers and employers on what AI skills are needed for hiring was particularly wide. Among job seekers, 37 percent said "verifying and correcting AI outputs" — catching errors in AI-generated content — was the most critical skill, followed by "workflow redesign," or using AI to improve how tasks are performed, at 18 percent. Yet only 15 percent and 3 percent of companies, respectively, said they required those skills.

In all, 55 percent of job seekers believed advanced AI capabilities were necessary, while only 18 percent of companies said the same — meaning job seekers felt roughly three times more pressure than employers actually expected.

Some 68 percent of companies said basic, practical-level AI skills were sufficient, and 14.5 percent said they did not require any AI competency at all upon entry — a stark contrast to the 0.6 percent of job seekers who perceived the same.

KEF said both the minimum expectations of companies and the current skill levels of job seekers were concentrated at the "basic use and task application" level, but that job seekers believed they needed to go further. This mismatch, KEF said, likely causes job seekers to feel more pressure in their preparation than companies actually demand.

The assessment of AI skills during hiring also remains in a transitional phase. Some 20 percent of companies said they formally include AI proficiency as an official evaluation criterion, while 43 percent said they consider it informally as a reference. Another 37 percent said they do not factor it into hiring at all. Among those that do assess it formally, 16 percent apply the criterion only to certain roles — the majority of that group.

Rather than becoming a universal hiring requirement across all positions, AI proficiency is being selectively assessed based on the nature of each role or used as supplementary reference material.

Looking ahead, companies rated "data interpretation" as the most important competency for new hires to develop as AI becomes more widespread, scoring it 73.9 out of 100. "Problem definition and problem-solving ability" followed at 73.2, with "critical thinking and judgment" close behind at 72.5.

Accordingly, companies identified "the ability to link AI with one's major or job function" (36.3 percent) and "problem-solving ability" (35.7 percent) as the areas universities should prioritize to strengthen new hires' AI competency — meaning the ability to apply AI in conjunction with specialized knowledge and to analyze problems and derive solutions using AI tools.

KEF advised job seekers to build foundational problem-solving skills rather than simply learning to use AI as a tool. It said they should practice analyzing data in line with the goals of a given task and verifying the validity of AI-generated outputs — capabilities that will allow them to adapt to new technologies and tools as they continue to emerge.

"This survey confirmed a significant gap between the level of AI competency job seekers think they need and what companies actually expect," said Kwon Hyeok-min, head of KEF's growth strategy division. "As the ability to interpret data and solve problems is expected to become increasingly important, university education must evolve beyond simply teaching AI technology toward practical, hands-on training that connects AI with students' majors and future roles."