90% of executives say AI hasn't boosted productivity. Some are still cutting jobs

TEXT START: Business leaders and investors face a deepening paradox: Companies are pouring more money into artificial intelligence than ever, but they’re not seeing the gains in productivity that they expect.

The Dissection

This article domesticates an existential claim. It accepts AI investment and relocates the failure from economic displacement to bad management: layoffs create fear, fear damages sentiment, and damaged sentiment suppresses productivity. Its prescription is to train workers, share gains, and preserve labor peace so firms can extract more value from AI.

That is a useful account of short-term implementation friction. It is not an account of the system’s trajectory. The article turns a structural conflict—AI becomes valuable partly by reducing dependence on labor—into a solvable workplace-culture problem.

The Core Fallacy

It confuses transition drag with terminal constraint. Employee insecurity can reduce present productivity, but it does not invalidate AI’s eventual cost and performance advantage over human cognitive labor. Positive sentiment cannot make redundant labor economically necessary.

A firm that retains workers to protect morale carries costs that more aggressive competitors can remove. “Share the gains” is therefore a temporary bargain, not a stable equilibrium. Training may improve human performance now while producing a more efficient workforce that can later be automated more completely.

The article also treats weak current productivity evidence as a challenge to AI-driven layoffs. That does not follow. Firms can cut in anticipation of future substitution, to free capital, or to satisfy investors before the technology has matured. Early layoffs can be strategically premature without being structurally irrational.

Hidden Assumptions

• AI’s main value comes from augmenting existing workers rather than eliminating tasks, roles, and occupations.
• Morale is the central bottleneck, rather than one temporary cost of reorganization.
• Firms can voluntarily distribute AI gains without competitors forcing labor costs downward.
• Upskilling creates durable human economic roles instead of new layers of automatable work.
• Executive perceptions, employee reviews, announcements, and stock reactions adequately establish the causal productivity mechanism.
• Layoff announcements attributed to AI reliably identify why the cuts occurred.
• The mass-employment, wage, and consumption circuit remains the system’s durable baseline.
• Firm profitability and individual worker viability ultimately point in the same direction.

Social Function

Partial truth serving as transition management and ideological anesthetic.

The article correctly identifies a real mechanism: fear can make workers resist tools that threaten them. But it offers managers a way to preserve the AI project by making displacement more acceptable, not a way to preserve mass productive participation. Its humane language masks the competitive pressure that eventually makes retaining economically replaceable labor irrational.

The Verdict

Operationally right, structurally evasive. AI layoffs can sabotage productivity when deployment is immature and the remaining workforce is demoralized. But the proposed cure—training, shared gains, and fewer immediate cuts—can improve AI adoption and accelerate the eventual severing of employment from income.

This is not a refutation of the Discontinuity Thesis. It is a manual for making obsolescence work more smoothly while the postwar labor circuit dies underneath it.