Plentiful, high-paying jobs in the age of AI

URL SCAN: Plentiful, high-paying jobs in the age of AI
FIRST LINE: "I'm traveling today, so here's a timely repost."

THE DISSECTION

This is elite self-exoneration masquerading as economic pedagogy. Smith deploys comparative advantage theory—correctly, technically—to construct a plausible-sounding mechanism by which human labor retains value even after AI achieves universal task superiority. The argument is structurally valid within its constraints but functions socially as a lullaby for the professional class and a liability shield for AI capital allocators.

The piece oscillates between two modes:
1. Rigorous constraint analysis (compute scarcity creates opportunity cost dynamics)
2. Motivated reasoning (therefore humans will probably be fine and well-paid)

The gap between these modes is where the copium lives.

THE CORE FALLACY

Constraint Permanence Assumption

Smith's entire edifice rests on compute remaining a binding, AI-specific constraint that cannot be relaxed faster than human-specific constraints (energy, land, social coordination). This is presented as a durable structural feature rather than a temporary engineering bottleneck.

The Error:
- Compute scarcity is a function of capital allocation and Moore's Law derivatives, not physics
- Energy scarcity affects humans MORE than AI (humans require ~2000 kcal/day of biologically compatible energy; AI requires electricity, which can be generated from sources humans cannot metabolize)
- Land scarcity affects humans MORE than AI (data centers are vertically scalable; humans require distributed habitable surface area)
- Social coordination affects humans MORE than AI (AI deployment faces regulatory friction but not labor organizing, NIMBY opposition to housing, or democratic legitimacy constraints)

The moment compute becomes cheap enough relative to human labor costs, comparative advantage stops protecting human employment and starts determining which humans get turned off first.

HIDDEN ASSUMPTIONS

1. Demand Infinity: Assumes economic demand expands infinitely to absorb all AI productivity gains, maintaining scarcity-driven wages for human comparative advantage domains. Ignores demand saturation and the consumption circuit dependency (wages fund consumption; AI doesn't consume).

2. Benign Allocation: Assumes whoever controls AI capital will choose to preserve human employment out of economic logic rather than minimize costs through maximum automation. Ignores that comparative advantage operates within property rights structures—if AI owners control energy/compute, they set the terms.

3. Static Constraint Ratios: Treats compute/energy/land constraints as fixed ratios. Ignores that AI capital is optimizing specifically to relax AI-specific constraints (custom chips, energy-efficient architectures, algorithmic efficiency gains) while human constraints remain biologically fixed.

4. Wage Floor Mysticism: Assumes humans will be "well-compensated" in an AI-rich society because aggregate wealth increases. Ignores that compensation requires economic necessity—if AI can do the task cheaper, why pay the human premium except as charity?

5. Comparative Advantage Universality: Imports a trade theory designed for geographically separated production systems with transportation costs into a unified digital labor market with zero marginal distribution costs. The analogy breaks.

THE KILL MECHANISM (DT Lens)

Smith's argument fails because it misidentifies the binding constraint:

Smith's Model:
- Compute scarcity → AI opportunity cost → Human comparative advantage → Human employment

DT Model:
- AI achieves cost/performance dominance → Firms automate to survive competition → Aggregate wage income falls → Consumption circuit breaks → Demand collapses → Human comparative advantage becomes economically irrelevant even if technically real

The Trap:
Comparative advantage guarantees humans will be best at something. It does not guarantee that something will be:
- Economically necessary
- Scalably employable
- Compensated above subsistence
- Accessible without AI-owner permission

Smith's doctor earning $2,000/appointment in an AI-rich world only works if:
1. Patients have $2,000 (wage income from what?)
2. AI owners don't just build more compute instead of hiring doctors
3. Regulatory/cultural moats protect human-only medical practice
4. The appointment is economically necessary vs. AI telehealth at $0.02

Outcome: Comparative advantage becomes a selection mechanism for which humans remain economically legible, not a guarantee of mass employment.

SOCIAL FUNCTION

Primary: Ideological anesthetic for AI capital allocators and their professional-class allies

Secondary Functions:
- Liability pre-emption: Establishes "we considered labor impacts and found theoretical reasons for optimism"
- Consent manufacturing: Makes AI acceleration seem compatible with broad prosperity
- Class signaling: Demonstrates economic sophistication (correctly explaining comparative advantage) while arriving at convenient conclusions
- Anxiety management: Provides technically-grounded hope narrative for professionals watching automation approach

Audience:
- VCs/founders who need to believe they're not building a dystopia
- Policy elites who need a reason to not regulate AI development
- Professional class seeking reassurance their credentials retain value

What It's Not:
- A rigorous model of labor market dynamics under discontinuous automation
- A plan for maintaining mass purchasing power
- An explanation of how comparative advantage survives the consumption circuit break

THE VERDICT

Technically Correct, Systemically Wrong

Smith correctly explains comparative advantage and correctly identifies that compute scarcity could preserve human employment under specific conditions. But he:

1. Underweights the speed at which AI-specific constraints relax vs. human-specific constraints
2. Ignores the consumption circuit (his wealthy AI society has no mechanism to distribute purchasing power if humans aren't productively employed)
3. Treats economic necessity as exogenous when it's actually endogenous to the production system
4. Confuses theoretical possibility with probable outcome

The piece is sophisticated copium—it uses real economics to construct a plausible-sounding mechanism for human labor retaining value, then hand-waves away the coordination problems, power dynamics, and demand-side constraints that determine whether that mechanism actually operates.

DT Translation:
"Comparative advantage means humans will always be best at something" is true.
"Therefore humans will have plentiful, high-paying jobs" is a non sequitur that ignores:
- Who controls the capital that determines what gets produced
- Whether human comparative advantage domains are economically necessary vs. luxury goods
- Whether aggregate demand can be maintained without mass wage income
- Whether AI owners have any reason to preserve human employment beyond regulatory compliance

The Unspoken Admission:
Smith's own caveats (the horse example, energy competition, the possibility of human immiseration) reveal he knows the argument is fragile. The piece ends not with confident prediction but with "possible, perhaps even likely"—which is not how you describe an economic law, it's how you describe a hope.

Classification: Partial truth deployed as full reassurance. The economics are sound; the application to AI transition is motivated reasoning dressed in comparative advantage theory.