Understanding AI Infrastructure:
A Map of Inquiry
Eight domains of inquiry, organized so that each conversation thread can stay focused without losing sight of the larger picture. Items marked ✓ Covered have been addressed in earlier sessions.
How to use this guide
Use this as a checklist before starting a new conversation thread. Pick a domain, identify where you left off, and begin there. Ask for a summary at any point to capture what's been established before the thread grows too long.
Domain Key
Physical / Infrastructure
Political / Regulatory
Technical / How It Works
Societal / Human Impact
Economic / Financial
01
Data Centers: Physical Reality
Physical
→Where are data centers being built — Pacific Northwest, California, nationally? ✓ Covered
→What are data centers actually used for — AI, cloud, surveillance, entertainment? ✓ Covered
→Water cooling: the problem and the alternatives ✓ Covered
→Power consumption: how much electricity does the AI buildout require?
→Land use: how large are these facilities, and what do they displace?
→Who owns what — Microsoft, Google, Amazon, Meta footprints?
→Undersea and modular data centers — experimental or viable?
→Waste heat reuse: capturing and redirecting thermal output
02
Energy & Environment
Physical
→What energy sources power data centers — coal, nuclear, hydro, renewables?
→How does AI power demand interact with regional power grids?
→Carbon footprint of training vs. running AI models
→Nuclear revival: are data centers driving new reactor investment?
→Hydropower dependency in the Pacific Northwest — vulnerabilities?
→What happens to power grids when AI demand spikes?
→Water table and watershed impacts of large-scale cooling operations
03
How AI Actually Works
Technical
→What is a large language model and how is it trained?
→Training vs. inference — what's the difference in resource terms?
→What are GPUs and why does AI require so many of them?
→What does it mean for a model to "hallucinate"?
→Open-source vs. proprietary AI — what's at stake?
→AI agents: what are they and how do they differ from chatbots?
→How do AI companies improve their models over time?
04
Surveillance, Privacy & Data
Societal
→What personal data is collected and stored — and by whom?
→How is behavioral data used for advertising and profiling?
→Government access to data center contents — legal frameworks
→Facial recognition and AI-enabled surveillance infrastructure
→What protections exist — GDPR, CCPA, federal gaps?
→China's AI surveillance model vs. Western approaches
→Biometric data: what's being collected without clear consent?
05
Politics, Policy & Regulation
Political
→Which government entities govern data center approval — local to federal? ✓ Covered
→What questions should voters ask candidates at each level of government? ✓ Covered
→Federal AI legislation — what exists, what's proposed?
→State-level AI laws — California, Washington, Oregon leading?
→Tax incentives for data center construction — who benefits?
→The EU AI Act — does it affect U.S. companies?
→AI and antitrust — are a few companies too dominant?
→Water rights law as applied to data centers
06
Labor, Economy & Displacement
Economic
→Which jobs are most at risk from AI automation — near term?
→Which jobs and industries are being created by AI?
→Data center construction: local employment vs. long-term local jobs
→Who profits from the AI buildout — concentration of wealth?
→Small businesses and AI: tool or threat?
→How much does it cost to train a frontier AI model?
→AI and healthcare costs — productivity vs. access
07
AI in Daily Life
Societal
→AI in healthcare: diagnostics, drug discovery, patient records
→AI and education: tutoring, cheating, institutional response
→AI in creative fields: art, music, writing — tool or replacement?
→AI and accessibility — benefits for people with disabilities
→AI and misinformation — deepfakes, synthetic media
→Social media algorithms as a form of AI — how they shape behavior
→AI companions and mental health — benefits and concerns
08
The Bigger Picture: Risk & Future
Societal
→What is AGI (Artificial General Intelligence) and how close is it?
→AI safety — what are researchers actually worried about?
→Concentration of AI power: a few companies vs. global access
→AI and democracy — election interference, propaganda at scale
→International AI race — U.S. vs. China, strategic implications
→What does a world with abundant AI look like in 10–20 years?
→Who decides how AI is governed — corporations, governments, or publics?
On Thread Strategy
Physical and environmental topics (Sections 01–02) work well together in one thread. Political and regulatory topics (05) can be woven in when they directly arise from a physical question, or kept separate for focused policy discussions. Sections 03 and 08 are good starting points for any conversation about the technology itself. At any point, ask for a summary to capture what's been established before starting a new session.