Book Note: Empire of AI Part 8 - The Gambit and Cloak-and-Dagger

Empire of AI Part 8 - The Gambit and Cloak-and-Dagger

This is part 8 of a nine-part reading of Karen Hao’s Empire of AI. Its scope is Chapter 15: The Gambit and Chapter 16: Cloak-and-Dagger. This public note does not reproduce the book. It uses chapter titles, the book’s public framing, and independent interpretation to build a reading lens.

The guiding question is: AI 경쟁에서 비공개 전략과 정보 비대칭은 공적 감독을 어떻게 어렵게 만드는가?

■ How to use this note

Book notes are storage; insight cards are currency. This note keeps claims, questions, and applications rather than long quotations.

L0 · Entry

• Core sentence: 경쟁의 속도가 빨라질수록 책임의 흔적은 더 느슨해지는 것이 아니라 더 촘촘해져야 한다.
• Why read this: I want to read the AI race as a question of power, resources, labor, and accountability, not only model capability.
• Author context: Karen Hao is a journalist who has reported extensively on AI industry and OpenAI.
• Scope: Chapter 15: The Gambit / Chapter 16: Cloak-and-Dagger

L1 · Captures

■ Copyright boundary

This public note does not reproduce long passages or continuous scenes. It offers chapter-level summary and independent interpretation only.

• Conceptual anchor: accountability
• Why it matters: It helps connect a technical promise to the power structure that makes the promise possible.
• Reader response: Any account of AI’s future should also name who bears the cost and who can object.

L2 · Chapter Map

#	Chapter	Reading question
1	Chapter 15: The Gambit	How do technical missions and social power become entangled?
2	Chapter 16: Cloak-and-Dagger	What does that entanglement change about resources, labor, and auditability?

Argument in one paragraph:

고위험 기술 경쟁은 비밀과 속도를 정당화하기 쉽다. 이 회차는 정당한 보안과 책임 회피를 구분하기 위한 최소 공개·감사·이의제기 조건을 묻는다. The goal is not to replace the book with a summary. It is to develop a reading lens that asks who decides, what is consumed, and what can be publicly checked.

L3 · Insight Index

• Empire of AI insight 8.1: promises must be read alongside their costs
• Empire of AI insight 8.2: AI evaluation must include conditions outside a benchmark
• Empire of AI insight 8.3: the ability to object is a feature of technology governance

L4 · Production Board

■ Output pipeline

• Korean draft: 승부수와 은밀한 작전
• Reviewed English draft: The Gambit and Cloak-and-Dagger
• Reading-note enrichment: add personal cases and exact page references after reading

L5 · Connections And Review

• Connection: AI governance, data-center infrastructure, compute resources, and human review.
• Action: When reading an AI proposal, record capability, cost, resources, accountability, and avenues for objection.
• Open questions:
  • AI 경쟁에서 비공개 전략과 정보 비대칭은 공적 감독을 어떻게 어렵게 만드는가?
  • What would need to be measured if this issue became a product or education decision table?
• Final takeaway: 경쟁의 속도가 빨라질수록 책임의 흔적은 더 느슨해지는 것이 아니라 더 촘촘해져야 한다.