Sapiens
Problem-Oriented Reading
Question 1: How did humans transform from an insignificant species to the masters of Earth?
Answer: At the collective level, humans can cooperate flexibly in large groups. Here, flexibility is relative to other species on Earth. For example, ant colonies can cooperate efficiently, but they would never plot to overthrow their queen. Large-scale cooperation represents a trust mechanism, a contract that “binds” strangers together.
Some cooperation mechanisms that only appear in the human species:
-
Slaughterhouses
-
Prisons
-
Concentration camps
Insight: The reason humans can engage in flexible large-scale cooperation is partly due to human self-awareness. But is artificial consciousness really the future of current AI development? It took millions of years for humans to evolve into a “conscious” species. Current computer interfaces don’t have the same ability as humans to interact with the real world and screen genes through natural selection. Therefore, AI might skip consciousness and become an unconscious superintelligence. Or perhaps computers will directly evolve super-consciousness through their unique interface, the “internet.”
Question 2: Why were humans able to establish unprecedented cooperation mechanisms?
Answer: Imagination. Humans can create fictional or imaginary stories. As long as people widely “believe”/“trust” these stories, everyone will follow the same rules, norms, and values.
- Animals only use language to describe the real world, while humans use language to create new (fictional) realities. For example: God, human rights, nations, ethnicities, companies, money.
Insight: The rapidly developing AI will have the same type of mechanism. For example, in the medical field, there are tens of millions of human doctors, but their varying levels of skill and cultural backgrounds often lead to misdiagnosis. AI doctors sharing a single medical network will share the same “standards” and can immediately access the latest drug-resistant viruses, which is undoubtedly enormous progress for healthcare.
2. Chronological Reading
-
Cognitive Revolution: Linguistic communication for cooperation.
-
Agricultural Revolution: Domestication of other species on Earth; use of writing for recording.
-
Scientific Revolution: Humans discovered their own “ignorance” and thus continuously explored unknown territories.
-
?? Revolution: ??
Insight: Humans will unconsciously enter a completely new era. During the transformation between eras, class mobility will become very rapid, and each transformation will concentrate resources in fewer hands. More and more power will flow from humans to algorithms, and most likely, an unconscious intelligence will control this planet.
Homo Deus
Opinion-Oriented Reading
Humans have always been pursuing “immortality,” while computers (silicon-based life) are naturally “immortal.”
Insight: According to evolutionary theory, carbon-based life completed evolution through continuous “death” to screen mutated genes suitable for survival. In contrast, silicon-based life doesn’t need this method to complete evolution. So do we need to encode a “primitive” silicon-based life form (a microcontroller with basic code fragments) to simulate the human evolutionary process?
The technological revolution produced humanist religion.
Science itself has become a religion. “The Bible” laid the foundation for theistic religions. “Newton and the apple,” “Markov assumption,” “neural networks” have become the cornerstones of computer science and science itself. I often fear that if one day the cornerstones of computer science collapse, this discipline will…
Humanism will face complete failure.
The basic assumption of humanism is that humans have self-awareness. But neuroscientists tell us that human consciousness is just neural signals, not much more sophisticated than other species on Earth, and there’s no 21-gram “soul.” In the future, dataism is very likely to become the new mainstream belief, where all low-level carbon-based life evolves together with high-level silicon-based life, ultimately becoming a huge integrated life form.
Chapter-by-Chapter Reading
-
Happiness: The essence of happiness is electrochemical reaction signals transmitted between neurons. Perhaps there really is a string of code that can transmit a “happiness matrix” within a neural network?
-
Intuition: Biology’s own algorithm (survival of the fittest) has ensured that organisms on Earth are at least locally optimal. The most important part of human decision-making algorithms, “intuition” or the “unconscious part,” needs to be cleverly encoded.
-
Consciousness: Consciousness is essentially individual regression and collective evolution. Free will itself is just an algorithm. So first, how do we give cold code initiative? Do we need to encode a most basic survival motivation for them? Or are there other methods?
-
Left and right brain inspiration: The currently most popular GAN (Generative Adversarial Network) model can be compared to the human left and right brain. Brain science experiments show that the left and right brain are originally two different decision-making systems connected through nerve fibers, ultimately making consistent decisions. Therefore, from neurons to deep neural networks to adversarial neural networks, humans are inadvertently following the “correct” path of bionics. Step by step realizing artificial intelligence. (GAN-related research papers can be added here)
Summary:
-
At this point in time in the world, whether it’s China’s advocacy for “economic globalization” or America’s advocacy for “America First,” both are telling a story, creating an illusion.
-
PyTorch and TensorFlow will ultimately become a super platform that standardizes everything about AI evolution. C language is like the physical laws of the real world, and underlying hardware is like entities in the real world.
-
Humans are simply algorithms. AI needs a “mass-energy equation” for this field.
