When AI does the thinking, what could possibly go wrong? - The Japan Times

TEXT START: I used to hate Bill Gates, co-founder of Microsoft, with a white-hot intensity.

THE DISSECTION

This passage is a polished alarm signal, not a structural analysis. It frames AI as a turbulent technological transition, praises Gates for identifying danger, and focuses on the disappearance of everyday friction—what it calls inefficiency, growing pains or life lessons. The author establishes credibility through a personal anecdote, then substitutes generalized concern for an account of mechanisms, ownership and distribution.

THE CORE FALLACY

It treats AI’s disruption as a governance problem: leaders must “get this evolution right.” Under Discontinuity Thesis logic, that is too mild. If cognitive automation achieves durable superiority, the central break is not turbulence but the severing of the mass employment → wage → consumption circuit. Better leadership may manage the wreckage; it cannot preserve mass productive participation unless control of AI capital changes.

The passage also romanticizes inefficiency as human development without asking whether humans remain economically necessary once machines perform the underlying cognitive work. Friction may teach people lessons, but it does not create a durable labor market.

HIDDEN ASSUMPTIONS

• Productive human participation will survive the transition in some broadly accessible form.
• Institutions can coordinate stable human-only economic domains at scale.
• “Leaders, experts and communities” possess enough agency to redirect the system.
• AI’s turbulence is temporary and will settle into a new social equilibrium.
• The main problem is inadequate preparation rather than concentrated ownership and power.
• Transfers or preserved consumption could substitute for the loss of economic agency without fundamentally altering the system.
• Human judgment, experience and inefficiency will remain valuable because they are human, rather than only where they remain indispensable to owners of automated systems.

SOCIAL FUNCTION

Partial truth wrapped in transition-management rhetoric and ideological anesthetic. It acknowledges danger, which is more honest than boosterism, but leaves the decisive questions untouched: who owns the systems, who controls access to their output, who becomes economically unnecessary, and what force could compel redistribution? By locating the solution in vaguely defined leaders and collective preparation, it converts a conflict over power into a problem of responsible administration.

THE VERDICT

The passage sees smoke but mislabels the fire as turbulence. It correctly detects that AI will produce historical instability, yet it does not confront the terminal mechanism: automation can preserve output and consumption while abolishing the majority’s productive role. This is a warning label attached to a machine that is dismantling the staircase beneath the laboring class.