Billionaire Marc Lore's Wonder uses an AI algorithm to determine who gets promoted | Fortune

TEXT START: Marc Lore has built companies by trusting the numbers.

The Dissection

The article is normalizing the conversion of management judgment into an automated ranking system. Peer ratings, written feedback, “value above replacement,” promotion timing, and taekwondo-style organizational levels are assembled into a machine-readable hierarchy. The same company is simultaneously automating kitchen labor at a reported rate of 500 bowls per hour versus 45 for a human worker.

This is not merely a fairness reform. It is a control architecture: quantify the worker, estimate replaceability, automate advancement decisions, and use exceptions as training data until human discretion becomes residual.

The Core Fallacy

The central error is treating quantification as objectivity. The AI does not discover intrinsic employee value; it operationalizes peer perceptions, managerial preferences, organizational design, and the chosen definition of “value above replacement.” A score can make bias less visible while making the system’s judgments harder to contest.

More importantly, even a genuinely fair promotion model does not preserve productive participation. It improves the efficiency of selecting winners inside a labor system that is simultaneously removing the need for large amounts of human labor. The promotion algorithm is a sorting instrument attached to the same machine that is shrinking the queue of economically necessary workers.

Hidden Assumptions

• That twelve coworkers produce truth rather than coordinated reputation management.
• That leadership, behavior, and performance can be reduced to stable measurable signals.
• That replacement difficulty measures strategic value rather than scarcity, politics, or poor documentation.
• That fewer human overrides mean the model is improving, rather than the organization surrendering judgment to it.
• That feeding disagreements back into the model removes bias instead of hardening the institution’s existing preferences.
• That transparency about compensation makes an automated hierarchy legitimate.
• That Wonder’s growth, funding, valuation, and expected IPO indicate durable social usefulness rather than capital’s confidence in scalable automation.
• That advancement remains the relevant goal when automation is reducing the number of human positions worth advancing into.

Social Function

Transition management, ideological anesthetic, and partial truth.

The partial truth is that standardization can constrain some arbitrary managerial bias. The anesthetic is presenting automated labor valuation as fairness while concealing its deeper function: making people legible, comparable, replaceable, and governable at scale. The article turns an early form of algorithmic workforce control into a business innovation story, while the kitchen automation figures expose the harsher reality beneath it.

The Verdict

Wonder is not demonstrating that AI makes the employment order fairer. It is demonstrating that the employment order can be administered with fewer human judgments before fewer humans are needed at all. The promotion model is hospice care for the wage hierarchy: cleaner metrics, tighter control, and a more efficient path toward productive participation collapse.