[Kim Dae-sik Column] The Secrets of the Universe AI Will Discover, Humans May Not Understand

⑪ The Era of Science Without Humans

In George Orwell's "Animal Farm," the pigs who overthrow the farm owner while shouting the slogan of "equality" become the new masters of the farm, proclaiming that "All animals are equal, but some animals are more equal than others." Similarly, we can think this way: "All animals try to understand the world, but some animals try to understand it more than others." That is us humans.

What is "understanding"? Beyond the textbook definition, it can be interpreted that the more we understand, the more accurately we can predict future events and make choices that better aid survival. From this perspective, all living beings on Earth constantly try to perceive and predict the world. So how can we predict the future most accurately? Fortunately, most of the future is an extension of the past. The better we remember the past, the higher the probability that we can predict the future more accurately.

Humans have brains larger than any other animal in this world relative to body size. Thanks to this, we can store and process more information than other animals. Just as a 100GB hard drive can store more information than a 1MB hard drive, humans begin to discover statistical patterns based on more and older past information that other animals cannot find. The sun always rises in the east (or rather, the place where the sun rises was named "east") and sets in the west. Summers are hot, and winters are cold. And plump zebras come to the riverside every morning to drink water. So, if you hide and wait by the riverside in advance, wouldn't it be easier to hunt zebras without running hard? Life becomes easier the moment you understand the repetitiveness and cyclical structure of the world.

However, not all phenomena in this world have cyclical structures. A delicious big fish that appears and disappears before your eyes, fruits falling from tall trees swaying in the wind, vines catching fire struck by lightning... Most natural phenomena do not repeat identically. Is there a way to predict non-repetitive phenomena? The moment we understand "causality," we can replace nature's repetitiveness with our own actions! Flint stones that replace unpredictable lightning, fruits that fall when you shake the tree without waiting for the wind, and fish that can be caught using bait.

Humanity began to replace nature's cyclical structures through its own actions. Since nature became a tool rather than fate, humans could become masters rather than slaves of nature. But that was only the beginning. Humans started applying cyclical structures and causality even to areas beyond direct experience. Humanity began to believe that rivers fall as showers from the sky, the sun is a giant fireball, and birds fly to another world humans cannot reach. Various natural phenomena gradually began to be shaped into forms imaginable by the brain, and perhaps this is how most superstitions, myths, and religions began.

In the 21st century, humanity no longer explains the world with superstition and myths. We have mathematics, science, and technology. Although faith and tradition have been replaced by mathematics and experiments, modern science is still just an extension of humanity's hundreds of thousands of years of effort to understand nature's repetitiveness and causality. So what is the future of science? Generative artificial intelligence that writes texts, draws pictures, and codes based on astronomical amounts of data created by humans.

If artificial intelligence can control telescopes and particle accelerators and even design new experiments directly, wouldn't it be natural to entrust science and research to AI? Future AI, surpassing human intelligence and memory, might discover new cyclical structures and causal relationships in nature that we didn't even know we didn't know. Then, is the future of science science without humans?

The brilliant secrets of the universe that future AI will uncover?no matter how much they are explained, we humans might never understand them. Just as ants will never understand quantum mechanics no matter how much it is explained.

Daesik Kim, Professor, Department of Electrical Engineering, KAIST