AI Won't Fix Broken Software. Better Thinking Might | Opinion - Newsweek

TEXT START: The biggest mistake we can make about AI is confusing more software with better software, or assuming that what’s in place today can be made better by simply adding AI.

THE DISSECTION

This is a managerial adaptation memo disguised as a systemic analysis. It correctly identifies patchwork AI, broken workflows, unreliable data, and software fatigue. It also admits that AI will reduce labor requirements. But it frames the transformation as a software-quality problem: integrate AI properly, preserve human judgment, and let small teams achieve more.

That diagnosis is locally accurate and globally evasive. The article discusses how to make automation useful without confronting who owns the automation, who captures the gains, and what happens when useful automation makes most cognitive labor economically unnecessary.

THE CORE FALLACY

The article confuses better software with better labor economics.

Under P1, well-designed AI does not weaken automation. It strengthens it. Complete, specialized systems that quietly remove administrative work are precisely the machinery that lets firms grow without hiring proportionally. The article’s celebrated outcome—tiny teams accomplishing what once required departments—is a direct mechanism of productive participation collapse under P3.

“Humans remain responsible” is not the same as humans remaining economically necessary. Legal accountability, relationship management, judgment, and creativity can remain nominally human while the number of humans required to perform them collapses. A single adviser serving thousands is not proof that advisers are protected; it is evidence that adviser productivity has been multiplied and headcount can be compressed.

The article also assumes that competitive pressure will permit stable human-only domains. Under P2, it will not. Once one firm can operate with fewer workers at lower cost, rivals must adopt comparable systems or die. Good implementation accelerates this pressure; bad implementation merely delays it.

HIDDEN ASSUMPTIONS

• Productivity gains will flow to workers rather than owners of AI capital.
• New businesses and new occupations will appear at sufficient scale and speed to absorb displaced labor.
• Demand will expand enough to offset labor substitution.
• Human judgment, relationships, creativity, and expertise are permanently non-substitutable rather than temporarily scarce.
• Human responsibility implies continued employment rather than residual liability and oversight.
• Commercial viability will produce social viability.
• Small-team entrepreneurship will create broad employment instead of highly productive firms with minimal staff.
• AI fatigue and implementation failures can materially restrain adoption once competitive survival is at stake.

SOCIAL FUNCTION

Primary classification: transition management and ideological anesthetic. Secondary classifications: partial truth and elite self-exoneration.

The article gives business leaders sensible operational advice and accurately warns against bolting AI onto defective processes. That is the partial truth. Its ideological function is to portray mass displacement as a matter of product design and managerial competence. If the transition becomes painful, the implied culprit is bad implementation—not the ownership structure of automated production or the disappearance of wage-based participation.

The phrase “humans remain responsible” functions as an institutional alibi. It preserves the appearance of human centrality after the economic necessity of most human labor has been removed.

THE VERDICT

A polished managerial lullaby with a sound local diagnosis and a false systemic prognosis. The article correctly sees that bad AI creates expensive, dangerous clutter. It fails to follow the more consequential implication: the better AI becomes at delivering complete services, scaling small teams, and removing friction, the more efficiently it severs the labor-to-wage circuit.

Better thinking may fix broken software. It does not fix the ownership problem, the coordination problem, or the collapse of productive participation. This is a guide for making automation profitable and socially tolerable—not a theory of how the majority remain economically viable after automation wins.