May 13, 2026 · Dr. Seungbin Yim

AI-Proof vs AI-Vulnerable:
5 Criteria That Decide

"Doctors are safe, accountants are at risk" — this kind of job classification is wrong.

AI doesn't replace "jobs." It replaces "tasks." Within the same profession, some people become more valuable while others become expendable. The differentiator isn't industry — it's task composition ratio.

The 5 Criteria Framework

Criterion 1: Repetitiveness

The higher the proportion of repeated patterns, the greater the risk. Writing similar reports every day, organizing similar data, responding to similar emails — these are first-priority targets for AI agents.

Red flag: "70% of my work is roughly the same every week"

Criterion 2: Codifiability

Can the judgment be explained as rules? Decisions that break down into "if condition A, then do B" are learnable by AI. Conversely, decisions entangled with context, politics, emotion, and relationships resist codification.

Red flag: "Anyone could follow my work manual"

Criterion 3: Physical Presence

Does the work require being physically present? Nursing, plumbing, electrical work, on-site inspection — these resist AI replacement for now. But "physical" plus "repetitive" tasks like warehouse picking are already being roboticized.

Green flag: "My job isn't only done behind a screen"

Criterion 4: Relational Complexity

Work requiring coordination, persuasion, and consensus among multiple stakeholders is hard for AI to replace. However, not all "relationship" work is safe — call center support is "relational" but pattern-matchable, and is already losing ground to AI chatbots.

Green flag: "I have open-ended conversations with no predetermined answers every day"

Criterion 5: Emergent Problem-Solving

The ability to define and solve unprecedented problems. AI excels at recombining existing patterns, but defining entirely new problems remains a human domain.

Green flag: "Much of what I do has never been done before"

Self-check: Evaluate your current work against these 5 criteria.

High repetitiveness + High codifiability + Low physical presence + Low relational complexity + Low emergence = Danger zone

If 3 or more apply, your career may already be in the "Quiet Erosion" zone.

The "I'm Creative, So I'm Safe" Delusion

After generative AI, the definition of "creative work" has shifted. Ad copy, design mockups, content planning drafts — previously labeled "creative," these are now produced by AI in 30 seconds, ten at a time.

Truly safe "creativity" is not producing outputs — it's deciding what should be produced. Strategic judgment, direction-setting, framing — these AI still cannot do.

What You Can Do Now

  1. Decompose your tasks: Log your work for a week by the hour and rate each task's AI-replacement potential from 0–100%.
  2. Expand your irreplaceable zone: Restructure your work to spend more time on the 20% AI can't touch.
  3. Use diagnostic tools: Replace vague anxiety with data-driven self-assessment.

Take the 3-Minute Assessment

10 questions. Instant results.
Find out where you stand against these 5 criteria.

Free Career Risk Quiz → Read "Quiet Erosion"