Steve Hanke Says AI Won't Replace Workers Because It's 'Incredibly Costly:' 'Businesses ...

URL SCAN: Steve Hanke Says AI Won't Replace Workers Because It's 'Incredibly Costly:' 'Businesses ...
FIRST LINE: Steve Hanke Says AI Won't Replace Workers Because It's 'Incredibly Costly:' 'Businesses Will Not Be Firing…'

The Dissection

The article converts a valid short-term constraint into a long-term forecast. AI requires expensive electricity, chips, water, data centers, and maintenance. That establishes friction and deployment lag—not immunity for human labor.

Hanke is measuring whether AI is free. Businesses are measuring whether AI is cheaper, faster, scalable, and more controllable than maintaining human labor over time. Those are different questions. The reported capital spending is not evidence that AI cannot replace workers; it is evidence that capital is being mobilized to build the replacement machinery.

The Core Fallacy

The central error is static-cost reasoning. AI does not need to be costless to displace workers. It only needs to become cheaper than the fully loaded cost of labor for specific tasks, then improve through scale, learning, and competitive investment.

The claim also assumes replacement must be immediate and total. It will not be. Businesses can remove hiring, compress teams, automate routine cognitive tasks, and increase the output expected from each remaining worker. That is sufficient to destroy labor’s bargaining power before mass layoffs become visually obvious.

Under the Discontinuity Thesis, this attacks the wage-to-consumption circuit incrementally. P1 does not require every job to vanish at once. P2 prevents institutions from preserving large human-only cognitive domains indefinitely. P3 follows when enough economically necessary work is performed by capital rather than labor.

Hidden Assumptions

• AI infrastructure costs will not fall rapidly through scale, specialization, and efficiency gains.
• Human labor costs consist only of wages, while AI costs include every visible infrastructure expense.
• Resource constraints will block deployment rather than create an investment race around energy, chips, cooling, and logistics.
• Firms will preserve labor-intensive processes even when competitors use automation to lower costs.
• “Not firing everybody” means employment remains structurally secure.
• Productivity gains will translate into broad wage gains rather than fewer workers producing more output.
• Physical bottlenecks can reverse the direction of technological substitution instead of merely delaying it.

Social Function

Classification: partial truth functioning as ideological anesthetic and transition lullaby.

The article gives incumbents a respectable reason to discount the threat: AI is expensive, therefore labor supposedly remains safe. The first clause is true. The conclusion does not follow. It reassures workers without addressing ownership, control, or distribution of the productive capital being built.

Its most useful contribution is identifying a real lag defense. Power, water, chips, permitting, and capital availability can slow the transition. They cannot restore the post-WWII employment bargain once AI becomes superior enough across economically important cognitive tasks.

The Verdict

Hanke has identified the price of the weapon and mistaken it for a prohibition on firing the army. AI need not replace everyone to make most workers economically unnecessary; it only needs to outperform labor at enough tasks while falling in cost. The article is therefore a near-term deployment observation dressed up as a systemic refutation—and the refutation fails.