The Bottleneck of Human Development
From steam engines and electricity to airplanes, wireless communication, rockets, and the Moon landing, humanity’s major breakthroughs have repeatedly transformed our way of life and our picture of the world. In an era when the total stock of knowledge was still relatively limited, a person could reach the scientific frontier of the time within a decade or so, and contribute across several fields: Newton did pioneering work in mathematics, physics, and optics; Poincaré — the mathematician who proposed the Poincaré conjecture — also left a deep mark on several branches of mathematics, including topology and dynamical systems.
The common foundation of these breakthroughs was that researchers could grasp problems from a global perspective and combine ideas from different fields.
Since the 19th century, and especially after the 20th, human knowledge has exploded: disciplines have kept splitting, new theories have emerged one after another, and the total has grown beyond what any one person can exhaust in a lifetime. Since then, breakthroughs have increasingly depended on division of labor and collaboration, and it has become hard for a single genius to complete a panoramic breakthrough from a global vantage point as before.
Given the current volume of human knowledge, an individual simply does not have enough time to master the full panorama, and a single brain can hardly hold such a vast amount of information at once.
The Opportunity AI Brings
Although it is still too early to draw conclusions, AI is very likely one of the most transformative technologies of recent decades. It promises to ease two human bottlenecks: lives that are too short, and brains that are not powerful enough. Given sufficient data, computing power, and electricity, AI can systematically learn a large body of digitized knowledge; its accumulation does not depend on any single person’s lifespan, so it does not have to start over when an individual departs.
Humanity therefore no longer has to wait only for the arrival of a “genius”; instead, through the accumulation of computing power, data, and algorithms, it can obtain scalable cognitive capacity. This capacity does not automatically grow stronger with time — it remains constrained by data quality, algorithmic bottlenecks, energy costs, and model obsolescence; but to a certain extent, it is becoming like electricity: something that can be produced and used at scale.
As the tool currently most promising for integrating knowledge across disciplines, AI may make global integration and breakthroughs possible again.
The Place of Humanity
Before AI, people called computers “thinking machines”, but strictly speaking, computers at that time were still essentially “calculating” machines, executing logic programmed by humans; software was human thought encoded, not thought intrinsic to the computer. Computers can be seen as the steam age of the “thinking machine”: they strengthened the execution of human logic, making it faster and more accurate. AI is closer to its electronic age: many of its rules of inference are not written line by line by people, but formed during training from large amounts of data.
Whether computers or AI, they are essentially tools for humans to achieve their purposes — like horse-drawn carriages and cars, they need people to set the direction. The difference is that AI is more autonomous, so human oversight and responsibility have not become lighter, but more important: tools are only the means; the ends are still decided by people.
In the next era, AI may serve as a container of human knowledge, a public thinking machine, partly freeing humanity from solving “how” so that we can focus more on answering “what” — and “why”. At that point, the question itself may be harder than the answer.