This is not only a developer test — CJ put an AI baseline across every role

CJ OliveNetworks announced its second-half 2026 graduate recruitment on September 17. Openings include AI, Data, software, DX, ERP, information security, infrastructure services, cloud, network services and AI video product planning.

The most notable change is the redesign of its previous coding-test process into an AI competency assessment for every role. Developers take a vibe-coding test in which they work with AI to build an output. Non-developers take an AI-use assessment focused on how effectively they apply AI in workplace situations.

The company describes the goal as hiring ‘AI Native’ talent. The public material does not say every applicant solves the same problem. The important structure is that the assessment format differs by role group while AI-use capability itself is checked across the recruitment population.

The bigger change is not ‘more AI tests’ — it is a change in the list of common capabilities

Role-specific expertise does not disappear. Developers still need to understand code, security professionals still need to assess risk, and planners still need to structure customer and business problems. There is no evidence that one AI score can replace those different disciplines.

But the meaning changes when a company begins checking the same type of capability across candidates for otherwise different professions. That can signal a shift in the organization-wide baseline rather than the addition of one more specialist skill inside a single department.

Banseog frames this as Role-specific Skill → Enterprise-wide Common Capability. A horizontal capability layer is added on top of the vertical expertise of each profession.

Growth in AI jobs and an AI baseline across all jobs are not the same phenomenon

JobKorea’s September 16 analysis of first-half 2026 postings found that positions registered by employers as AI specialist roles increased 128% year over year. Non-development AI roles rose 325%, while their share of AI specialist postings increased from 12% to 22%.

Growth was especially strong in roles such as AI content creator, AI planning, AI business strategy and AI education or consulting. That is evidence that the boundary of specialist work connected to AI is expanding beyond software development.

CJ OliveNetworks raises a different question. It is not only how many jobs will carry AI in the title, but what common operating capability will be expected even from roles that do not. Expansion of AI occupations and a shift in the baseline for all occupations should be analyzed separately.

For candidates, the answer is not simply to switch into an AI job

It would be an overreach to conclude that everyone must become an AI specialist. A more practical change is that candidates may increasingly need to preserve their domain expertise while showing that they can perform that work effectively with AI.

A marketer still needs to understand markets and customers. Finance still requires numerical control and validation. Security still requires knowledge of risk and systems. AI can change the speed and method of research, comparison, analysis and implementation without replacing the domain itself.

That makes ‘I have used ChatGPT’ weak evidence. Stronger evidence is an example of solving a real problem in the target function with AI, catching errors in the output, and explaining why the final answer was chosen. The vertical axis is domain expertise; the horizontal axis is the ability to control tools and convert them into outcomes.

Employers cannot stop at one common test — good AI use differs by profession

Assessing a common capability across roles is not the same as giving everyone the same test. Useful AI behavior in software development can differ from useful behavior in planning, security or infrastructure.

That is why the split in CJ OliveNetworks’ design matters: vibe coding for developers and workplace-scenario AI use for non-developers. A capability can be common while the observable behavior that demonstrates it remains role-specific.

The next design problem for employers is not simply whether a candidate used AI. It is translating each profession into criteria for where AI improves outcomes, which errors the candidate must detect, and where final judgment and accountability must remain with the human.

BANSEOG VIEW | The jobs stay specialized. The baseline becomes horizontal.

Calling this simply ‘more AI hiring’ misses the more interesting change. Developer and non-developer work remains different, and specialist knowledge does not vanish. But if different professions begin sharing a new baseline capability, the architecture of hiring changes.

The jobs stay specialized. The baseline becomes horizontal. Professions can remain deeply differentiated while a new common layer sits beneath or above them across the organization.

One company does not establish a Korean labor-market standard. Still, an explicit AI competency assessment across all roles is a clear signal that hiring is moving beyond the question of who builds AI toward another question: who can do their own job well with AI.

BANSEOG VIEW

Banseog View — The jobs stay specialized. The baseline becomes horizontal.

CJ OliveNetworks applies an AI competency assessment across all roles in its 2026 second-half graduate recruitment while using different formats for developer and non-developer candidates.

Banseog separates this from the growth of AI occupations themselves. The vertical axis of role expertise remains, but a horizontal enterprise-wide capability layer may be emerging across professions.

Candidates may benefit more from proving they can solve and verify problems in their own profession with AI than from trying to become generic AI specialists. Employers, meanwhile, must translate a shared capability into role-specific observable behavior.

SOURCES

Primary sources and references

CJ OliveNetworks has not publicly disclosed the detailed scoring rubric, test questions or passing threshold. JobKorea figures describe changes in AI-related and specialist postings on its platform and are not generalized here as statistics for the entire Korean labor market. ‘Role-specific Skill → Enterprise-wide Common Capability’ and ‘The jobs stay specialized. The baseline becomes horizontal.’ are Banseog interpretations based on the public evidence.