Apollo Chief Economist Says There's No Evidence AI Is Taking Human Jobs

TEXT ANALYSIS PROTOCOL

TEXT START: "Anyone worried that AI will replace them should take a deep breath, at least according to Apollo Global Management's chief economist."

1. THE DISSECTION

This is a staged reassurance artifact — a document whose primary function is legibility management for capital interests during an active structural rupture. Sløk is not analyzing the economy; he is narrating the experience of capital during the early phases of displacement, using lagging indicators (current payroll numbers, hiring of AI-adjacent roles) as evidence that the machine isn't eating. The Business Insider framing — amplification of executive optimism through peer agreement (Levie, Dell, Sacks, Goldman) — is textbook prestige laundering. The "zero evidence" claim is a statistical sleight of hand: evidence exists in real-time corporate disclosures, executive memos, and announced headcount reductions, but these are dismissed as isolated "lazy narratives" (Huang) or "AI washing" (Altman). The result is a perfect circular defense where actual evidence of displacement is redefined as either deception or irrelevant.

2. THE CORE FALLACY

Lagging indicator fetishism. Sløk's entire evidentiary base — April payroll additions of ~110,000, current hiring of AI experts, data center construction — operates entirely within the lag window. The Discontinuity Thesis holds that this window is precisely where displacement is masked: new AI capital formation creates short-term demand for implementation labor, hardware deployment, and integration work, all of which register as employment growth even as the underlying productivity logic is eliminating the workforce those systems will replace. Citing the ADP report as evidence that AI isn't displacing workers is equivalent to citing construction employment during a building's demolition phase as proof that no building is being destroyed — while the wrecking ball is being assembled. The Jevons Paradox framing is particularly grotesque here: Jevons described coal consumption increasing as efficiency rose, but that describes demand for the input resource, not participation in its production. The displaced workers Sløk's framework erases are not the AI experts being hired — they are the clerical, analytical, coordination, and operational labor whose functions are being absorbed. Jevons would be horrified at this usage.

3. HIDDEN ASSUMPTIONS

• Assumption 1: Human labor as a homogeneous category. Sløk implicitly treats "human workers" as a single pool, then cites the creation of a narrow AI specialist tranche as evidence that the entire pool benefits. Under DT logic, the relevant metric is not aggregate headcount but productive participation rate — and the displacement of median-skill knowledge workers is already measurable in disclosed corporate actions.
• Assumption 2: Current hiring patterns predict structural employment capacity. The AI investment boom is a transition event, not a terminal equilibrium. Every AI implementation expert hired today is being trained to build systems that will eventually render implementation experts unnecessary. Sløk is reading the construction phase of automation as proof that the automated factory will need workers.
• Assumption 3: Corporate headcount is the correct unit of analysis. Block going from 10,000 to 6,000 while maintaining output is not a neutral data point — it is a direct measurement of productivity displacement. Dorsey's memo is not spin; it is an honest accounting of the mechanism Sløk denies exists.
• Assumption 4: CEO sentiment surveys (60% of EY financial service CEOs) constitute evidence. This is testimonial evidence from parties with direct financial and reputational incentives to minimize displacement narratives. That 40% of surveyed CEOs did not predict headcount maintenance should be the headline, not the reassurance.

4. SOCIAL FUNCTION

Classifications: Elite self-exoneration, ideological anesthetic, transition management.

This article serves a specific institutional function: it provides a plausible-deniability vehicle for capital class actors to publicly minimize displacement while internally accelerating automation. The mechanism is identical across the cited agree-ers (Levie, Dell, Sacks, Goldman) — each benefits from a labor market that remains quiescent during the transition. The Dorsey/Cisco/Atlassian/IBM disclosures are real, but they are reframed as exceptions that prove the rule, or dismissed as "lazy narratives." The article functions as transition management: it creates a media record of expert disagreement with displacement claims that can be cited when political or regulatory friction emerges. It is not economics. It is reputation insurance.

5. THE VERDICT

Sløk is diagnosing the patient by examining the hospital's construction budget during active surgery. The evidence he cites — current payroll additions, AI hiring, Jevons paradox — is precisely the lag-phase evidence that the Discontinuity Thesis explicitly identifies as masking the underlying mechanism. The direct corporate disclosures he dismisses (Block, Cisco, IBM, Coinbase, Snap, Atlassian, Cloudflare) are not anecdotes; they are the first cohort of a cascade. His framework cannot account for why companies would simultaneously build AI systems and publicly announce workforce reductions citing those systems unless it redefines those companies as liars — which is itself a form of denial. The Jevons paradox is a category error: Jevons described more coal being burned because it became cheaper; Sløk is describing fewer humans being employed because AI becomes cheaper. These are opposite outcomes. Calling this "Jevons paradox in real time" is not economics — it is rebranding the collapse as a party.

Verdict: The article is a textbook example of institutional lag denial — using transitional demand signals as proof that structural displacement is not occurring. The data it dismisses is more probative than the data it cites.