An AI-native competency probe is a structured assessment of how a candidate collaborates with AI on a practical task — scoring the process (problem define → instruct AI → gather evidence → verify) across axes such as Technique, Intent, and Cognition — before or instead of treating the final artifact as hire-time proof. Proof they can do the job is a role-specific task whose output a hiring manager can open, read, and defend in a debrief (a work sample or job simulation). A competency probe answers “how well do they work with AI under shared conditions?” It does not by itself prove “this person can do this job’s day-one deliverable on our JD.”
That distinction went mainstream on 13–14 Oct 2026, when Korean coverage (MoneyToday, Seoul Economic Daily, Aju Press) reported Krafton’s largest new-graduate drive: the 2027 Hire-Track / recruitment-linked internship, aiming for more than 100 AI-native hires across six tracks (AI research, tech, support, business, game programming, game planning). Applications opened 13 Oct through 23 Oct 2026; tests and interviews run November–December, with results in January 2027 and a February start. The AI competency step uses Krafton’s in-house solution Cofa-Probe (also rendered “Copa Probe” in some English coverage; product site: Cofa Probe) — developed by Krafton’s AI Frontier team and previously field-tested at the July Kopaton / Cofathon hackathon with CJ Olive Young. Cofa-Probe scores Technique / Intent / Cognition from outputs plus prompts, tool use, verification traces, and post-task Q&A — not final deliverables alone. Krafton has also said it will invest about ₩30 billion a year on employee AI tooling. Schmidt and Hunter’s classic meta-analysis (Psychological Bulletin, 1998) still puts work-sample tests near .54 predictive validity for job performance — an AI-collaboration process score is closer to “how they steer the model” than to openable JD-native work the hiring manager can defend.
This guide is for TA leaders and hiring managers who hear “they cleared Cofa-Probe / high Technique–Intent–Cognition / AI-native competency pass” and wonder whether they still need a job task before panel time. It separates AI-collaboration process screens from hire-time proof, and shows where Dotof AI Role-Simulation Assessments fit: after (or instead of) competency probes, prove day-one work before you spend scarce interviews. Related (not rewrites): AI-native graduate hiring vs work sample, AI fluency interview vs work sample, AI interviewer vs job simulation, digital-twin interview vs work sample, and mission interview vs work sample.
What does an AI-native competency probe actually deliver?
It delivers a scored read of the candidate’s AI-collaboration process under a shared task, workspace, and model — not a score on Candidate A’s day-one output for Req X. In Krafton / Cofa’s framing, Probe collects artifacts, code, prompts, work logs, tool settings, and verification records; then maps them to Technique (implementation quality, security, verification), Intent (clear goals, constraints, and role handoff to the model), and Cognition (explaining how results work, spotting risks, recovering from failures). Post-task Q&A checks explanations against the actual implementation. Outputs are screening artifacts: competency scores, evidence packs, and interviewer follow-up questions.
Treat the probe as a process screen for “can they steer AI responsibly?” Celebrate when it surfaces who verifies vs who rubber-stamps the model. Do not import “high Cognition score / cleared Cofa-Probe” into a new-hire debrief as proof that Candidate A can ship this job’s day-one deliverable — especially when the probe task is a shared generic workspace rather than your live JD.
Why isn’t a Technique / Intent / Cognition score enough at hire time?
Because the hiring manager’s job in a debrief is to open evidence of work on this req — not a rubric of how someone collaborated with a model on a standardized AI-native task. A competency probe answers “how did they define problems, instruct AI, and verify?” A work sample answers “what would this person produce next week on our JD?” Managers defend shortlists to peers with the second answer. A candidate can score well on Intent and Cognition and still leave the interview packet empty of role output. When the only artifact is a process scorecard, the panel rediscovers capability live — which is the expensive screen the probe was never meant to replace.
The shelf also mixes different products under one “we assess AI natives” story. AI-native graduate programs thicken who you recruit into. AI fluency bars thicken how candidates talk about tools. Conversational AI interviewers thicken talk-first screens of the live person. Digital twins thicken interactive Q&A with a dossier proxy. Competency probes thicken scored AI-collaboration process. All can be useful context. None answer “can they do the work?” until someone opens a role-specific sample.
When should HR use a work sample instead of (or after) a competency probe?
- Use competency probes for AI-collaboration triage when the org is explicitly hiring “AI natives” (Krafton’s 100+ Hire-Track scale) and the bottleneck is distinguishing who verifies AI output vs who ships unchecked — and only after you can name a human who still meets the person before offer.
- Add proof before panel time once the role has observable day-one deliverables, a wrong hire is costly, or managers refuse to interview without something they can open. Generate the task from the live JD — see from job description to role simulation.
- Prefer the sample over the probe score as the hire gate for professional roles where “can they do the work?” is the question you cannot afford to get wrong. Use the probe to clear AI-process risk; use the sample to decide who advances to offer. Process scores help reach — they still do not produce openable job output for your req.
Do not ask the simulation to answer pure process-rubric questions (“what was their Intent score on Cofa-Probe?”) — keep that with the probe report and a named human. Keep a written rule: no Technique / Intent / Cognition score should be the only hire vote. The practical gates are in the responsible AI hiring checklist and the integrity cousins in prevent AI cheating in hiring assessments.
Which should we buy for skills-based hiring in 2026 — an AI-native competency probe or a work sample?
Buy the sample for the hire gate — and treat Cofa-Probe-style competency probes as optional process infrastructure that still needs openable proof underneath. A competency probe is a screening purchase: scored AI-collaboration process under shared conditions. A work sample is a selection purchase: openable evidence before panel hours. If your buyer question is “what should TA use instead of ‘they cleared the AI-native probe’ as the hire gate?”, start with the artifact managers will open in a debrief — then decide whether a Technique / Intent / Cognition report is still useful for funnel quality. That is the answer to “AI native competency probe vs job simulation”: score the process if you must; prove the work either way. See also AI interviewer vs job simulation, AI screening vs human judgment, and the anatomy of a useful AI assessment report for hiring managers.
How does Dotof fit this decision?
Dotof AI Role-Simulation Assessments start from the job description, generate a realistic task, run it in a supervised browser, and return structured evidence hiring managers can compare. Use a competency probe when your pain is “we cannot tell who actually verifies AI work.” Use the simulation when the decision you cannot afford to get wrong is whether this candidate can do the job — especially when the “assessment” so far was a process scorecard on a shared AI-native workspace. Process scores and job-task proof can coexist — they should not be confused.