Alex Im다는 and Phil Trammell – What remains scarce after AGI?

In one sentence

The economy will transition into an "Electricity Model" where AGI becomes a universal substrate, leaving human-centric "relational goods"—driven by evolutionary preferences for empathy and connection—and ownership of the AI index as the primary sources of scarcity.

Why read this (and who it's for)

This summary solves the paralyzing anxiety of "the white-collar apocalypse" by providing a rigorous economic framework for how automation actually reshapes markets. It is for investors, policymakers, and professionals who want to understand:

1. Why mass unemployment might be an overstated narrative.
2. How "structural change" creates new sectors of value.
3. How to position oneself (or a nation) in an era where software and manufacturing reach near-zero marginal costs.

The big idea

The central thesis argues that we are moving toward a Machine-only Economy for production, while maintaining a Human Economy for consumption. As AGI automates the supply chains of capital goods (software, manufacturing, research), those goods will reach 100% "network-adjusted capital share." This doesn't mean wealth disappears; it means the location of value shifts.

Because humans have intrinsic, evolutionarily ingrained preferences for human involvement—what the authors call the Relational Sector—the value of a barista, a ballerina, or a doctor providing empathy will remain high even when their technical tasks are automated. The "scarcity" of the future isn't "doing work"; it is "human connection" and "owning the index" of the machines that do the rest.

Key ideas

The Relational Sector

• What it is — Services where a human being "in the loop" is a core component of the product's value (e.g., a barista, a ballerina, or a doctor providing diagnosis). Value here stems from intrinsic preferences for empathy and connection rather than just scarcity.
• Why it matters — It identifies the "safe haven" of human economic value. Even if an AI can perform a task perfectly, humans may still prefer a human version because of evolutionary moral emotions.
• In the book — The authors cite Mark Zuckerberg hiring MMA instructors and dancers for his wife's birthday as an illustration of a preference for "relational goods" despite having massive wealth in non-relational assets (Meta stock).

Economics of Structural Change & The Lump of Labor Fallacy

• What it is — The historical reality that automation makes existing goods cheaper, which increases consumer spending power, which then flows into new sectors (like services).
• Why it matters — It debunks the fear that "automation = fewer jobs." Historically, technology expands the frontier; it is unlikely AI will replace engineers while remaining as expensive as human labor because that scenario lacks an "abundance effect."
• In the book — David Ricardo (1820) predicted mass unemployment due to automation. He was right about jobs being automated but wrong about employment rates because he missed how automation creates new jobs by making goods cheaper.

The O-ring Model of Jobs

• What it is — A job consists of various tasks (e.g., a doctor seeing patients vs. filling out insurance docs). Automation can target specific tasks within a job.
• Why it matters — If AI automates 9/10 tasks but the human focuses on the remaining task, productivity increases and prices drop. If demand is elastic, this leads to more hiring of humans for that final critical "O-ring" task.
• In the book — The authors note that while junior developer jobs are down, senior engineer demand is up, suggesting a shift in task composition rather than total job erasure.

Jevons Paradox & Elasticity of Software

• What it is — As a resource gets cheaper, total consumption increases (e.g., coal in Britain). This occurs only if demand is highly elastic.
• Why it matters — Because software is expected to be highly elastic, as it becomes cheaper via AGI, its usage/demand will skyrocket rather than plateauing.
• In the book — The authors argue that negative economic growth is improbable because the technological frontier is expanding; for growth to be negative, demand would need a hard bound (consumers saying "I've had enough"), which is unlikely for software.

Evolutionary Preferences & Moral Emotion

• What it is — Humans have evolved "moral emotions" against offloading social interactions to AI.
• Why it matters — Natural selection may point toward an even stronger preference for other humans over time, ensuring the Relational Sector remains a high-value niche.
• In the book — The authors reference Jonathan Haidt’s framework on moral emotions to explain why we value human connection regardless of machine capability.

The "Greedy Optimizer" Model

• What it is — Wealthiest individuals often possess a preference for accelerating capital rather than immediate consumption (a "Nick Landian preference").
• Why it matters — If an agent does not satiate quickly in capital, they will maintain a higher savings rate and eventually own the majority of the wealth and capital share.
• In the book — Elon Musk’s investments in both humans and machines while reproducing quickly is used as an example of this dynamic.

The "Electricity" Model of AGI

• What it is — The economy will run on AGI the way it currently runs on electricity—a broad, fundamental transformation where every S&P 500 company must leverage AI to remain indexed.
• Why it matters — If AGI is commoditized like electricity, it leads to "broad increases in prosperity" because gains are distributed through society rather than concentrated in a few labs.
• In the book — The authors argue that once AGI is commoditized, everyone will have access to the resource quickly via open models.

Frameworks & tools

Relational Sector Analysis

Use this to determine where human value will persist:

1. Identify Tasks: Break a job down into its constituent tasks (e.g., "Diagnosis," "Scheduling," "Insurance Filing").
2. Assess Human Preference: Ask: "If you don't care about humans being involved in that process, why would humans be involved?"
3. Categorize: If the value is intrinsic/relational (empathy, trust, physical performance), it belongs to the Relational Sector. If it is purely functional, it will likely reach 100% capital share.

The "Buy the Index" Strategy (For Developing Nations)

A developmental strategy for countries not currently in the production chain:

1. Avoid Naive Advice: Do not focus solely on retraining programs or building local data centers (which are hard to scale).
2. Identify the Substrate: Recognize that AGI is becoming the "Electricity" of the economy.
3. Purchase Ownership: Focus on sovereign wealth funds or subsidies that allow citizens to purchase a portion of the AI-driven surplus/index.

Memorable examples & stories

• The Phone Operator Precedent (1920–1940): Technology existed for ~20 years before full automation occurred. Workers were reabsorbed into the economy at lower salaries and remained "underemployed." Lesson: Automation is often a slow transition, not a sudden cliff.
• The Art Print Experiment: Consumers paid significantly more for a human-made art print than an AI version when it was unique; however, if 500 human versions were produced, the price dropped significantly because the "connection" factor was lost. Lesson: Scarcity is tied to the uniqueness of the human connection.
• Zuckerberg's Birthday Party: Zuckerberg hiring MMA instructors and dancers for his wife's birthday illustrates a preference for "relational goods" despite having massive wealth in non-relational assets (Meta stock). Lesson: Human intrinsic goods remain valuable regardless of how much capital you own.
• Nigeria & Mobile Banking: Nigeria skipped middle developmental steps for "astronomical growth," similar to how mobile banking became more prevalent in Nigeria than in Germany. Lesson: AGI allows developing countries to "leapfrog" previous industrial stages by buying the index.

Key quotes

• "If you don't care about humans being involved in that process, why would humans be involved in that process?" Why it matters: This defines the boundary of the Relational Sector.
• "The more you think AGI is going to be, our economy is going to be run on AGI the way our economy currently runs on electricity." Why it matters: Shifts the perspective from AGI as a "tool" to AGI as a "utility/substrate."
• "It's more difficult to imagine something that doesn't exist that's a good thing than losing something that exists." Why it matters: Explains why negative narratives dominate despite the economic evidence of structural change.
• "Nigeria just needs to buy the index." Why it matters: A simple, powerful strategy for global equity in an automated world.

How to apply it

This week

• Audit your tasks: Take your current job and list every task you perform. Mark which ones are "functional" (could be done by a robot) and which are "relational" (require human empathy/trust).
• Identify the "O-ring": Determine which of your relational tasks is the most critical for your employer's success. Focus your professional development there.

Ongoing

• Focus on Relational Goods: In your career, prioritize building skills that involve moral emotions (trust, empathy, coaching) over purely functional technical skills that are easily commoditized.
• Seek Ownership of the Index: As an investor or citizen, look for ways to own shares in the AGI infrastructure rather than just "using" the software.

Common pitfalls

• Narrative Bias: It is easier to imagine losing a job (something that exists) than imagining a world where AGI creates new types of prosperity (something that doesn't exist yet).
• Ignoring Structural Change: Many fear mass unemployment because they forget how automation historically makes goods cheaper, which frees up capital for new sectors.
• The "White Collar Apocalypse" Fallacy: This assumes that AI will replace software engineers while remaining nearly as expensive as human labor; the authors argue this is unlikely because it lacks an "abundance effect."

Connects to

• Jevons Paradox: Explains why cheaper resources (like software) lead to higher consumption rather than lower usage.
• Moral Emotions (Jonathan Haidt): Provides the evolutionary framework for why humans will always prefer human interaction in certain sectors.

If you remember only three things

• The economy is shifting toward an Electricity Model where AGI becomes a universal, commoditized substrate.
• Value will concentrate in the Relational Sector (human-centric goods) and Ownership Shares of the AGI infrastructure.
• Structural Change ensures that as capital goods reach 100% automation, consumer spending power shifts into new human-centric sectors.

Internalize it

• Reflection questions

  • Which parts of my daily work would I still want a human to do even if an AI could do it perfectly?
  • If software becomes essentially free, what new "relational" services will emerge as a result of that freed-up capital?
  • How can I position my personal portfolio (or my country's economy) to own the "index" of AGI rather than just being a consumer of it?

• Flashcards Q: What is the "Relational Sector"? A: Services where human involvement is a core value component, driven by intrinsic preferences for empathy and connection. Q: What is the "Electricity Model" of AGI? A: The idea that AGI will become a broad, fundamental utility that every company must leverage to remain indexed. Q: Why did David Ricardo's prediction of mass unemployment fail? A: He missed the "economics of structural change"—how automation makes goods cheaper and frees up spending for new sectors. Q: What is the O-ring Model of Jobs? A: A job is a series of tasks; if AI automates most tasks, the human's value is tied to the remaining critical task(s). Q: Why is negative economic growth improbable in an AGI world? A: Because the technological frontier is expanding, and for negative growth, demand would need a hard bound (which is unlikely).

Fidelity note

Confidence: High. This summary captures the core economic arguments regarding the Relational Sector, Structural Change, and the Electricity Model provided in the notes. The specific "moral emotion" framework is attributed to Haidt as per the text. No outside information about current AI models was included.