Are You Good at AI? Companies Are Now Basing Performance Reviews on It

TEXT START: Bosses everywhere are scrambling to figure out if and how to grade their employees' AI use.

THE DISSECTION

This is an early symptom report, not a diagnosis of the disease. Corporations are turning AI from an optional tool into a performance criterion because they need behavioral compliance before they have redesigned work. They tried crude proxies—tokens, logins, and leaderboards—then retreated when those metrics exposed themselves as gameable and discriminatory. They are replacing them with softer proxies: speed, quality, impact, anecdotes, and teaching.

The article accurately records workslop, security risks, managerial confusion, subjective scoring, and the failure to redesign workflows. But it treats these as implementation problems. The deeper event is that firms are forcing workers to discover use cases, validate outputs, absorb risk, and prove their own productivity gains while management captures the resulting labor reduction.

THE CORE FALLACY

The text treats AI competence as a durable individual performance trait inside an essentially intact employment system. Under the Discontinuity Thesis, P1 makes cognitive automation increasingly superior; P2 prevents institutions from preserving stable human-only economic domains; P3 removes the majority from economically necessary labor.

The decisive question is not which employee uses AI best. It is how many employees the same output now requires. The best amplifiers may simply become the engineers of their coworkers' redundancy. Once AI fluency spreads, it stops being a moat and becomes the minimum admission price for temporary usefulness. Better reviews may select survivors during the transition; they cannot restore mass productive participation or preserve the post-WWII wage-consumption circuit.

HIDDEN ASSUMPTIONS

• AI-generated productivity will produce better careers and higher wages rather than lower headcount and greater concentration.
• Managers can reliably distinguish useful AI adoption from performative activity, despite the article's own evidence that measurement remains subjective and uneven.
• Workers have equal time, training, access, and freedom to experiment, regardless of job type, medical leave, parental leave, or security constraints.
• AI skill will remain scarce long enough to function as a durable career moat.
• Firms will eventually redesign workflows and career paths instead of merely intensifying quotas and surveillance.
• More output will map to more employment rather than proving that fewer humans are required.
• The corporation will remain a viable distribution mechanism after employment loses its productive function.

SOCIAL FUNCTION

Classification: transition management and partial truth, with an ideological-anesthetic effect.

The article gives workers a vocabulary for surviving the new ranking system and gives executives a script for demanding adoption before they understand the destination. It makes a system-level collapse look like a problem of individual fluency, initiative, and adaptation. Its warnings are real, but its frame keeps the reader inside the office-performance-review theater, where the question is how to score the worker rather than whether the worker remains economically necessary.

THE VERDICT

AI-based performance reviews are the paperwork of labor-market triage. Logins and token counts are primitive; impact, quality, and knowledge-sharing are more refined proxies, but they still identify who can convert human labor into fewer required humans.

The article captures the lag phase: confused metrics, workslop, security exposure, institutional inertia, and managerial improvisation. It does not reveal a path to system survival. Under DT, the performance review is being retrofitted from a development ritual into a rationing mechanism: prove that you can multiply yourself, or be classified as excess. The old economic order is not adapting safely. It is teaching workers to participate in their own replacement.