AI Makes Polished Work Cheap—Judgment Becomes the Rare Asset

The quality of a deliverable no longer reveals the quality of the thinking behind it. That shift, subtle but fundamental, is forcing a reckoning in how organizations identify and develop talent.

Shannon McKeen, writing in Forbes, describes a telling classroom experiment that illustrates the problem. In a business practicum course, students tackled a foundational consulting exercise: building an issue tree to break down a messy client problem into structured questions and hypotheses. Half the students used AI assistance.

The AI-assisted submissions were polished—clean logic, tight structure, plausible hypotheses. The work produced without AI looked messier, with rougher edges and some muddled thinking. Graded on the artifact alone, the AI-assisted work earned higher marks.

Then came the presentations. The pattern reversed completely. Students who had relied on AI to structure their thinking stumbled when questioned, offering vague and off-target responses. Students who had wrestled with the problem themselves, even clumsily, demonstrated clear understanding and could defend their reasoning. The AI had improved the artifact without improving comprehension.

Why it matters

This gap extends far beyond the classroom. For decades, knowledge work has been evaluated through proxies—the sharp memo, the clean financial model, the tight presentation deck. These artifacts were difficult to produce, so producing them signaled genuine capability. AI has severed that connection. A polished report now indicates access to a capable model, not necessarily a capable thinker. Every hiring process, performance review, and promotion decision built on evaluating outputs is suddenly measuring something it can no longer see.

What remains scarce

When AI makes competent-looking work abundant, what becomes valuable is what the technology cannot replicate: judgment under uncertainty.

A model can draft stakeholder analyses but cannot identify which stakeholder will quietly derail a project. It can generate defensible options but cannot tell you which one to stake your reputation on when data is incomplete and timelines are tight. It can produce recommendations but cannot be held accountable for them.

Judgment is built through friction, consequence, and the accumulated experience of being wrong in ways that matter. Organizations that route cognitive work through AI for short-term efficiency gains risk long-term cognitive atrophy. Research examining how generative AI affects critical thinking reflects this concern—the productivity gains appear immediately, while the erosion of judgment shows up later.

The experiential advantage

The solution is not using less AI but creating situations where AI cannot do the part that counts—and where shallow thinking becomes immediately visible.

McKeen observed that live, high-stakes moments reveal what polished deliverables hide. When students had to defend recommendations to actual clients or revise plans after stakeholder pushback, the question shifted from "Did AI write this?" to "Walk me through your reasoning." The live context became the assessment.

AI used well can sharpen rather than eliminate struggle. Teams that interviewed AI personas representing difficult stakeholders learned how the same recommendation lands differently with different audiences. Coaching AI that pushed teams to surface assumptions before conflict made them personal raised the difficulty rather than removing it. This distinction matters: AI that eliminates struggle erodes judgment; AI that intensifies productive struggle builds it.

Implications for hiring and development

Organizations must rebuild their evaluation systems around demonstrated reasoning rather than polished outputs. That means interviewing for decisions, not deliverables—putting candidates in front of ambiguous problems and observing how they reason aloud. It means protecting the work where struggle builds capability rather than offloading all cognitive friction. And it means valuing experiences that test people against real ambiguity, real pushback, and real consequences.

The details were first reported by Shannon McKeen in Forbes. The fundamental insight holds: AI has not made expertise obsolete, but it has revealed what expertise was always standing in for—judgment under uncertainty, exercised by someone willing to be accountable for the outcome.