Philosophy & AI
Why the Race to Artificial General Intelligence Ignores What Matters
The United States and China are locked in a race to achieve Artificial General Intelligence. Each assumes that arriving first will deliver supremacy over the other. Both are wrong — not because AGI is impossible, but because the race itself is based on a false premise: that intelligence, unmoored from consciousness, is the prize worth winning.
For the past century, the United States and China have maintained complementary economies. American innovation in design, finance, and intellectual property has been matched by Chinese manufacturing, infrastructure, and scale. The advance of one side has consistently benefited the other. This is not idealism; it is economic history.
Yet the race for Artificial General Intelligence proceeds as if this complementarity does not exist. Egged on by military establishments on both sides, the competition frames AGI as a zero-sum game: whoever achieves machine intelligence that matches or exceeds human cognition will dominate the other. The focus is on security, not prosperity. On weapons, not welfare.
This is the logic of the Cold War applied to a technology that bears no resemblance to nuclear weapons. A nuclear bomb destroys. A generally intelligent machine, if it arrives, would build — but what it builds depends entirely on who directs it, and toward what ends.
The AGI race resembles nothing so much as the competition between Jeff Bezos and Elon Musk to be first to reach Mars — as if civilization on Earth were already behind them. Two billionaires pouring resources into escape velocity while the planet they inhabit faces crises of water, heat, inequality, and displacement. The AGI race is the geopolitical equivalent: two superpowers pouring hundreds of billions into machine intelligence while four billion human beings in the Global South remain trapped in informal economies, without access to the AI tools that could transform their lives.
The competition is not for AGI. It is for the market that is truly up for grabs: four billion citizens in the lower-middle class who could be uplifted by a combination of wise politics and wise markets.
As the arms race accelerates, four geopolitical questions of enormous consequence are being treated as externalities — inconveniences to be managed rather than imperatives to be addressed.
Question One: How Does AI Affect Global Warming and the Distribution of Water? Training a single large language model can consume as much electricity as a small city uses in a year. Data centers require staggering quantities of water for cooling — millions of gallons daily. The AGI race, by definition, demands ever-larger models, ever-more computation, ever-greater energy and water consumption. Meanwhile, the nations most vulnerable to climate change and water scarcity are in the Global South — the same populations the race ignores.
Question Two: How Does AI Affect Inequality? Every major advance in AI capability so far has concentrated wealth upward. The gains flow to those who own the models, the data, and the infrastructure. AGI, if achieved, would be the most powerful wealth-concentrating technology in human history — capable of replacing not just manual labor but cognitive labor across every sector simultaneously.
Question Three: How Does AI Serve the Global South? AI is being developed about the Global South — as a market to capture — but not for the Global South, as populations to empower. The four billion citizens of the lower-middle class across Asia, Latin America, and Africa represent the largest untapped market in human history. Yet neither the American nor the Chinese AI ecosystem is designed to serve them.
Question Four: How Does AI Affect Materialism? This is the question nobody in the AGI race is asking, and it is the most important of the four. The race is driven by a materialist assumption: that intelligence is computation, that more computation equals more intelligence, and that more intelligence equals more value. It never asks what intelligence is for.
For all of human history, intelligence and consciousness have been joined. To think was to be aware. To reason was to experience. Every act of intelligence carried with it the weight of meaning, suffering, joy, mortality. Human intelligence was never pure computation; it was computation saturated with consciousness.
Artificial intelligence breaks this bond. For the first time, we have intelligence without consciousness — vast computational power that can analyze, predict, generate, and optimize without experiencing anything at all. The machine does not suffer when it makes a decision that causes suffering. It does not rejoice when it creates beauty. It does not die, and therefore it does not value life.
The core principle is this: consciousness must direct intelligence, not the reverse. This is not a Buddhist teaching alone. It is central to the teachings of Islam and Christianity. It is the universal insight of every contemplative tradition that has observed the human mind.
Jensen Huang, the CEO of NVIDIA, has given his name to a law of computing acceleration: Huang's Law, which observes that AI computing power is now advancing faster than Moore's Law ever predicted. Machine intelligence doubles and redoubles. Human consciousness does not. The gap between what machines can compute and what humans can comprehend widens with each generation of chips.
The term "prompt engineering" has entered common usage to describe the skill of directing AI systems through carefully constructed instructions. It is a useful term, but it is the wrong one. Engineering implies a mechanical skill — the optimization of inputs to maximize outputs. What is actually required is something far deeper. I propose the term prompt science.
Why science and not engineering? Because science, since Aristotle, has been tied to the pursuit of truth. The Greek word episteme — from which we derive "epistemology" — meant not just knowledge but justified true belief. Physics — the foundational science — asks what the universe is made of and how it behaves. It does not merely build machines; it seeks to understand reality.
The practitioner of prompt science must cultivate two capacities simultaneously: technical fluency with AI systems and self-knowledge — awareness of one's own consciousness. This is where fifty centuries of contemplative tradition become not a cultural artifact but an operational necessity. The meditator, the practitioner of mindfulness, the person who has trained their attention — these individuals are better prompt scientists, because they understand the instrument that directs the machine: their own mind.
Prompt science is the discipline by which consciousness maintains command over intelligence. It is the practice that prevents the decoupling from becoming permanent.
The AI Middle Way Coalition does not compete with the AGI race. It renders the AGI race irrelevant.
Whether AGI arrives in 2027 or 2035 or never, four billion people in the lower-middle class still need AI-enabled tools now: cooperative formalization that brings informal economies into productive participation, multilingual interfaces that serve indigenous languages, microcredit systems that expand access to capital, educational platforms that build human capability. These are not AGI problems. They are governance problems, market problems, and problems of political will.
Kenneth Boulding, the great systems theorist, identified three systems of power: the threat system (power through destruction), the exchange system (power through trade), and the integrative system (power through shared identity and values). The AGI race operates entirely within the threat system. The AI Middle Way operates within the exchange and integrative systems — building shared frameworks that create value for all participants.
The race to AGI is a race to Mars. The AI Middle Way is a commitment to Earth — to the four billion people who live here, who are conscious, who suffer, and who deserve intelligence in the service of their flourishing.
The Bangkok Declaration, to be signed on April 21, 2026, formalizes this commitment. It is not a declaration against AGI. It is a declaration for consciousness — for the principle that intelligence, however powerful it becomes, must serve the aims of aware, living beings.
The superpowers may reach AGI. They may not. Either way, the four billion are waiting. And consciousness — patient, aware, irreducible — will still be here when the race is over.
Craig Warren Smith is Chairman of the Digital Divide Institute and Co-Director of the AI Middle Way Coalition. A former visiting professor at Peking University and Harvard Kennedy School, his "Meaningful Broadband" framework was adopted as national policy by Thailand (2006) and Indonesia (2007). He has practiced Buddhist meditation since 1971 and teaches Mudra Space Awareness, a contemplative discipline rooted in Tibetan Buddhist monastic tradition.