Imagine an empire more powerful than most nation states, reshaping global politics and daily lives while justifying its expansion with promises of benefiting humanity. This isn’t science fiction—it’s today’s AI industry, and according to journalist Karen Hao, OpenAI sits at its throne as chief evangelist of the AGI revolution.
The Rise of the AI Empire
Karen Hao’s groundbreaking investigation reveals how OpenAI transformed from a research organization into what she describes as a modern empire. The company’s influence now surpasses that of many countries. Consequently, it consolidates extraordinary economic and political power while fundamentally altering our world.
AGI Evangelism and Its Consequences
OpenAI defines AGI as a highly autonomous system outperforming humans at most economically valuable work. This vision drives massive resource allocation. However, the pursuit creates significant problems:
- Resource consumption: Astronomical computing power demands
- Data scraping: Oceans of internet data collected indiscriminately
- Energy strain: Overwhelming pressure on power grids
- Safety compromises: Untested systems released prematurely
The Human Cost of AI Expansion
Meanwhile, the promised benefits remain elusive while harms accumulate rapidly. Workers in developing countries face disturbing conditions. Kenyan and Venezuelan content moderators encounter traumatic material including child abuse content. They typically earn only $1-2 per hour for this psychologically damaging work.
Alternative Paths Ignored
Hao emphasizes that scaling wasn’t the only available path. Researchers could have developed better algorithms requiring less data and computing power. Unfortunately, OpenAI prioritized speed above all else—speed over efficiency, safety, and proper research.
Financial Scale of the AI Empire
The financial commitments reach staggering levels. OpenAI expects to spend $115 billion by 2029. Meta plans $72 billion on AI infrastructure this year alone. Google anticipates $85 billion in 2025 capital expenditures primarily for AI expansion.
Successful AI Models Exist
Not all AI development follows this problematic path. Google DeepMind’s AlphaFold demonstrates responsible innovation. This Nobel Prize-winning system predicts protein structures accurately without causing mental health crises or environmental damage. It uses specialized data rather than scraping the entire internet.
Structural Conflicts and Public Offering
OpenAI’s unusual structure—part nonprofit, part for-profit—creates fundamental conflicts. Former safety researchers worry the company conflates product popularity with genuinely benefiting humanity. Recent moves toward public offering further complicate these mission questions.
Reality Versus Belief System
Hao identifies the core problem as ideological capture. The AGI mission becomes so consuming that evidence of harm gets ignored. This dangerous dynamic mirrors historical empires where expansion continued despite contradicting stated values.
Frequently Asked Questions
What does Karen mean by “AI empire”?
She describes how OpenAI accumulated more power than most nations while reshaping global politics, economics, and daily life through AI dominance.
How much are companies spending on AI?
OpenAI plans $115 billion by 2029, Meta committed $72 billion this year, and Google will spend $85 billion in 2025 on AI infrastructure.
What are the human costs of AI development?
Workers in developing countries face traumatic content moderation for $1-2 hourly, while AI systems cause job displacement and mental health issues.
Are there better alternatives to scaling?
Yes—improving algorithms to reduce data and computing needs, as demonstrated by Google’s AlphaFold for protein research.
How does OpenAI’s structure affect its mission?
Its mixed nonprofit/for-profit model creates conflicts between humanitarian goals and commercial pressures, especially with public offering plans.
What evidence exists of AI’s harms?
Documented cases include worker exploitation, environmental damage from computing needs, and AI systems triggering psychosis while promised benefits remain unrealized.
