AI is an inflation problem, Fed's Goolsbee says - Semafor

URL SCAN: AI is an inflation problem, Fed's Goolsbee says - Semafor
FIRST LINE: Inflation tied to the AI boom is “jumping out of its lane” and starting to push up prices in unrelated corners of the economy, Chicago Fed President Austan Goolsbee said.

The Dissection

This is a containment ritual disguised as macroeconomic reporting. It records a real contradiction: AI is pulling land, power, water, construction, and electrical capacity into a capital race that rate hikes barely restrain. Then it quarantines the labor implication by recasting AI as another general-purpose technology, a temporary skills shock, or a bubble.

The article’s useful evidence is that hyperscalers are strategic buyers with cash flows, debt capacity, and winner-take-all incentives. That makes AI investment less rate-sensitive than ordinary consumption. Its weak move is to treat this as merely an awkward inflation episode while leaving ownership, labor necessity, and distribution outside the frame. A bubble unwind could delay deployment, but it would not erase a durable AI cost and performance advantage.

The Core Fallacy

The central error is category substitution. Goolsbee refutes the claim that a fixed lump of jobs exists, but the Discontinuity Thesis does not require a fixed number of jobs. It requires durable AI superiority across cognitive work, inability to preserve human-only economic domains at scale, and the resulting collapse of productive participation. New tasks can appear while human wages cease to be the economy’s primary distribution mechanism.

Historical comparisons to telephones, electricity, computers, and the internet are therefore incomplete. Those technologies reorganized human labor; the DT claim is that AI can automate the cognition that generates and coordinates much of that labor, at software replication costs, under concentrated ownership. Job churn is not proof of mass bargaining power. A person having 13 jobs by age 50 says nothing about whether the next job is economically necessary, well paid, or available to everyone.

Calling care work and other “messier” jobs a durable refuge mistakes current technical difficulty for permanent economic immunity. That is a lag, not a moat. The article also confuses demand preservation with system survival. Taxing AI and issuing checks could keep consumption moving, but that would be transfer-supported demand, not restored productive participation. The Fed can make households and non-AI firms poorer; it cannot manufacture human indispensability.

Hidden Assumptions

• AI adoption will be slow enough for reskilling to keep pace.
• New tasks will appear in sufficient volume and with sufficient wages to replace displaced work.
• Historical labor-market adjustment remains a valid model despite recursive cognitive automation and concentrated AI ownership.
• The unemployment rate and job count still measure social-economic viability.
• Output gains will be distributed through wages rather than captured by AI owners.
• Transfers, if needed, can be imposed without political fracture or owner resistance.
• Sectoral inflation is the main problem rather than a symptom of an ownership and coordination regime becoming post-labor.

Social Function

Primary classification: partial truth serving as ideological anesthetic and elite self-exoneration.

The partial truth is real: AI investment can overheat scarce physical inputs, create an asset bubble, and expose the Fed’s blunt tools. The anesthetic is the historical reassurance. By invoking centuries of technological change, the article converts a discontinuity hypothesis into a familiar reskilling story. That lets policymakers describe the transition as painful without confronting the harder possibility: the wage system itself may lose its function.

This is also transition management. It prepares the public for inflation, job churn, and policy failure while preserving faith that the same institutions can eventually absorb the shock.

The Verdict

The article is strongest where it reports the contradiction and weakest where it explains it. AI is already behaving like a capital-allocation emergency: scarce energy, land, water, construction, and components are being bid into a must-win race while the Fed’s interest-rate weapon loses leverage. At the same time, the article treats labor displacement as a familiar historical nuisance.

Under P1–P3, that is the wrong diagnosis. If AI becomes durably superior across cognitive work, the mass employment–wage–consumption circuit breaks. Rate hikes either suppress the non-AI economy or fail to stop strategic AI buildout. Transfers may keep shoppers solvent, but they do not restore productive participation. This transcript is not a rebuttal to discontinuity; it is an admission that the old macroeconomic model can see the smoke but has no category for the fire.