Everything starts with intelligence
The intelligence described in this passage refers to artificial intelligence (AI) created through the simulation of biological intelligence in humans and animals. Intelligence is a cognitive ability that can be accumulated through learning. Its function is to solve problems faced by intelligent entities. It is a form of logical ability to shape and solve problems, and learning (training) shapes and enhances intelligence.
In humans and animals, intelligence originates from the brain organ composed of protein structures. The neural network in the brain generates intelligence. In the brains of animals, neural networks can be found, but the structure and complexity of these networks cannot be compared to humans. Let’s take the human brain as an example and briefly understand how the protein neural network works.
The brain is composed of brain cells (Neurons). The distinguishing feature of brain cells compared to other cells in the body is that they have axons and dendrites. Brain cells connect with each other through the contact between axons and dendrites., as follows:
Axons are extensions of brain cells, and they are long and thin in shape. The shortest axons can be only a few millimetres long, while the longest ones can exceed a meter in length. These long axons are used to connect cells on the other side of the brain, as well as different parts of cells in different locations of the brain.
Neural impulses are composed of weak electrical currents. When this current travels from one end of a brain cell to the other, the starting end of the axon receives the electrical signal and releases a chemical substance called “neurotransmitter” through a process known as chemical reaction. This substance permeates through the gap between two brain cells and contacts the axon of the next cell, triggering the next neural impulse. This process repeats as the impulse travels to the other end of that brain cell, stimulating the secretion of neurotransmitter once again. The transmission of neural signals basically occurs through this relay of brain cells. Axons act as signal initiators and propagators, while dendrites can be understood as signal receptors. The combined effort of axons and dendrites completes the entire relay process.
The above description pertains to the micro-level transmission process. On a macro scale, each cell may simultaneously connect to thousands of other cells. This means that in an instant, a cell can receive thousands of neural signals from other cells and transmit them to another cell. We can imagine the network structure of the entire brain cells as having a three-dimensional shape. Due to the varying lengths of axons, brain cells are not only locally connected in a network fashion, but cells in different regions also have contacts across regions. The contacts between brain cells will diverge outward in a geometric progression, where one cell may connect to thousands of other cells, and each of those cells may further connect to thousands of subsequent cells. The brain consists of approximately 900 million such neurons in this network, a quantity comparable to the number of planets in the universe. This vast network structure is unparalleled by any artificial non-neural network system today, including computers or telecommunication networks. Its information processing speed is awe-inspiring.
From the perspective of computer science, the connection points between cells can be regarded as “logic gates.” The state of these logic gates can be changed or increased, and their connections with other gates can also be increased. By altering or increasing the quantity of logic gates through information input, the output logic result can be changed. This is a form of “training” for the neural network, which, from the perspective of neural networks, can be seen as “learning.” Neural networks are considered intelligent because they are not rigid organisations that execute instructions in a fixed manner (where a fixed output is obtained based on a given initial input condition).
The function of an intelligent entity is to process information, which involves two aspects: inputting and outputting information. Input refers to the introduction of information from the external world into the intelligent entity, while output refers to the feedback of information from the intelligent entity back to the external world. The output information can be referred to as the result, thus forming an information loop where information can be exchanged. The point of interaction between intelligence and the external world is called an “interface” (organs such as eyes, ears, hands, feet, and skin serve as interfaces). We can say that the intelligent entity inputs and outputs information through interfaces.
People and animals
Before the emergence of artificial intelligence, there were only two types of intelligent beings on Earth: human intelligence and animal intelligence. Ultimately, human intelligence triumphed over animal intelligence and became the dominant force on the planet. Now, let’s first understand the decisive role that “intelligence” and “interface” play in the competition between humans and animals, both being biological intelligent beings.
Since the extinction of dinosaurs, humans began to appear and gradually evolved into “Homo erectus.” From that moment on, humans became the “natural predators” of other terrestrial animals, and the competition between the two began until humans completely dominated the Earth.
If we overlook the factors of God mentioned in religion and the hypothetical intervention of extraterrestrial high-intelligent beings in the birth of humans, this article analyzes from the perspective of scientific evolution. Since humans and animals are both intelligent beings enveloped in interfaces, why is it that only humans continue to evolve and eventually become the highly intelligent beings we are today, while other animals, for the most part, remain at the level of biological structural changes without any improvement in intelligence? To explain this, we need to understand the facilitating role of the human interface in intelligence, which we refer to as “training/learning.
As mentioned earlier, both humans and animals possess a neural network structure made up of proteins. The complexity of the neural network is determined by the number of connections between neurons, which in turn influences the level of intelligence. The brain serves as the physical carrier of the neural network, and it is evident that a larger brain volume can accommodate more neurons. A larger brain volume is a requirement for a more advanced neural network. Furthermore, an increase in the number of connections and density within the neural network results in a higher brain weight. These conclusions have been confirmed in animals.
For instance, due to the buoyancy provided by water, whales have developed the largest and heaviest bodies in the animal kingdom, and they have been found to possess the highest intelligence among animals. In contrast, the brain volume of the Australian tree kangaroo is only the size of its eyeball, which explains why it moves slowly and can only focus on one task at a time. When it is eating, it cannot engage in other activities, and when it is moving, it cannot eat. The low-intelligence brain of the tree kangaroo can only handle very basic information processing.
We now understand the influence of brain volume on intelligence. Additionally, we recognise that humans have the ability to live upright. The vertical alignment of the spine enables it to counteract the force of gravity, allowing the spine to support a greater brain weight compared to horizontal-spine organisms. Therefore, the human brain has the potential for significant development in terms of both volume and the spatial and complexity of the neural systems within it. This sets the stage for future training and learning, as the human brain is ready to receive input and serve as material for training and learning.
The input of information requires an interface. Intelligent entities need interfaces to exchange information with the external world, and interfaces establish the pathway for information flow. The functionality and form of the interface play a crucial role in determining the quality and quantity of input information. It is precisely because humans possess interfaces with more powerful capabilities than animals that they can access higher quality and larger quantities of information. Through this, they have been able to train and develop their high-level intelligence, thus conquering the animal kingdom. So how did humans achieve this?
Firstly, humans evolved to walk upright on two feet, with eyes that possess fine vision and hands with ten fingers. The ability to walk on two feet gives humans unparalleled mobility, which is of great importance (the importance of mobility for improving or training intelligence will be emphasised multiple times in the text). Based on this mobility, an intelligent entity can move freely in different spaces and, more importantly, bring itself to the desired space to achieve the purpose of learning. In different spaces, the intelligent entity can gather external information through another interface – the eyes. Human feet not only allow for running on flat ground but also enable climbing steep terrains with the assistance of hands. This ability allows humans to traverse mountains and difficult terrains. With the assistance of upright legs and hands (for climbing and swimming), humans have dispersed to every corner of the Earth through long-distance migrations from the birthplace in Africa. This mobility of migration is unparalleled by other animals. Wherever humans migrated, the large-bodied animals became extinct. This is because humans have higher intelligence than animals. It is because, during spatial movement, the intelligent entity (neural network) can receive more information (and this information is what intelligence seeks). It is an exchange of information that achieves the goal of learning. It is through the ability of the intelligent entity to purposefully enter the forest, observe the desired behaviour of an insect, and achieve this kind of learning that only the mobility and visual capabilities of humans with the ability to receive detailed information can accomplish.
Another interface is the fingers, which serve as a bridge for interacting with the external world and obtaining information. In the realm of terrestrial organisms, only primates and humans have naturally developed fingers as tools. While chimpanzees have fingers similar to humans, they have not developed intelligence like humans, lacking sufficient intelligence to utilise their fingers effectively. As a result, they cannot achieve complete information feedback and, therefore, the purpose of learning. Humans, on the other hand, use their fingers for learning, basing their learning on acquired knowledge and using their fingers to create tools. In the process of using tools, they observe and summarise information, which becomes new knowledge. For example, human intelligence, through visual observation, discovers that round objects can be moved by rolling. With their dexterous fingers, humans can select different objects for experimentation, strengthening the cognitive abilities of their neural networks. Eventually, they may discover that cutting a cylindrical stick into a round shape allows it to be placed on the ground and objects can be moved on top of it with ease. This is the prototype of a wheel-based vehicle.
In this process from learning to tool-making, human unique interfaces play a crucial role: mobile feet, dexterous fingers, and eyes with fine vision. Additionally, the most important interface is the expandable neural network – the brain. Within the brain, the growth of the neural network allows for the recording of learning outcomes. This accumulation of knowledge is the essence of intelligence.
Via the interface, information is input from the external world into the neural network, which then outputs messages and changes the state of the external environment, leading to new feedback messages. This closed loop of information flow constitutes the interaction between the external world and the intelligent agent, which is the process of learning or training. The natural interfaces of intelligent agents have limitations, but humans have learned to extend their interfaces through toolmaking and domestication of other animals, breaking free from the constraints of natural interfaces.
Humans may not have the sharp fangs and claws of predators, but they have developed tools, including projectiles such as arrows. Humans may not possess great running speed, but they have tamed horses to pull carts or serve as mounts. Humans may lack the ability to traverse water, but they have learned to build ships. Humans may not have powerful muscles, but they can domesticate cattle and horses to transport objects and provide labor. As a result, humans have far surpassed animals both in terms of intelligence and in the extension of their natural interfaces. From then on, humans have become the masters of the terrestrial world.
People and people
The competition between intelligent entities is never-ending, and once humans conquer animals, the next intelligent opponent can only be themselves. The future competition will involve groups of intelligent entities competing against another group, leading to a clash of civilisations. Up until now, when we talk about intelligent entities, whether humans or animals, they are all biological beings composed of protein-based neural networks and interfaces, which I refer to as “protein-based intelligent agents.”
Protein-based neural networks have several characteristics. Firstly, one major advantage is their lightweight and high density compared to artificially manufactured chips (artificial neural networks are layered, with neurons distributed on a plane, where neurons are only connected to neighbouring neurons on the same plane, and there are no connections between different layers. This is fundamentally different from the three-dimensional fully connected neural networks in the brain, allowing protein-based neural networks to achieve higher density under current technological conditions). This high-density, lightweight, and small volume distribution of cells provides “mobility” to protein-based intelligent agents. Mobility is crucial because it allows intelligent agents to access a greater amount of diverse information. However, with advantages come disadvantages, and the biggest drawback of protein-based neural networks is the lack of permanent and accurate memory. The memories stored in the brain are easily “forgotten” over time, and when life ends, memories are permanently lost. Moreover, memories are often inaccurate, and we frequently experience “false memories” or errors in recalling information, even simple things like remembering a phone number can be prone to mistakes. It can be imagined that if a group of intelligent entities can compensate for this significant limitation, the civilisation formed by that group can excel in competition.
The solution is the emergence of writing. Writing uses external media, rather than proteins themselves, to store information. The recorded information is transformed into language or conceptualised text through visual images. Once readable information is recorded in a medium, its lifespan becomes synchronous with that of the medium. Different media have their advantages, and the development of media has progressed from stone and bamboo slips to parchment and eventually to paper. This development has aimed for low economic cost, high information density, and lightweight high-capacity media. At this stage, the drawbacks of protein-based memory media, such as short lifespan and information inaccuracy, are overcome.
Understanding the crucial role of mobility in the learning process of intelligent entities, written media (books) now offer portability, which is a form of reverse mobility. Additionally, it provides temporal mobility along the timeline. Without writing, if an intelligent entity in location A wants to access information from location B, it would need to physically move to B during its lifetime. However, with the emergence of writing, if an intelligent entity in location A can see written media (books) imported from location B, it no longer needs to travel for information and can save time. Moreover, within its lifetime, it can read information imported from locations C, D, E, and even F. More importantly, this is the first time that information can transcend timelines, allowing intelligent entities to receive information written by previous generations (intelligent entities). The birth of writing sets high requirements for maintaining the accuracy of information across different eras and regions, and not every script in the world can achieve this. Ancient Chinese characters, Latin, and Ancient Greek scripts reached this level. From then on, the range of information that intelligent entities can access expands across time and space. The quality and quantity of information determine the effectiveness of learning, and learning determines the intelligence of individual intelligent entities. Intelligent entities then record the summarised information from learning and output it in written form to the next time and space. The cross-temporal and cross-spatial nature of writing connects intelligent entities, liberating neural networks from individual intelligent entities and linking them with neural networks in another time and space through writing. This is the formation of intelligent groups. At this point, the value of intelligence is no longer limited to individual entities but is enhanced through more efficient group dynamics. This is the emergence of civilisation, and competition begins to take place between civilisations.
This is the well-known history of advanced civilisations which are the civilisations with more tools and stronger organisational capabilities—conquering land-bound. In an era before tools were produced in a modern scientific manner (extensions of interfaces), the use of animals or even plants played a significant role in advancing civilisations. For example, cattle and horses served as tools for transportation and pulling (European colonisers had domesticated cattle and horses, while Native American and Inca civilisations in the Americas lacked these important production and combat tools). The versatility of bamboo as a tool (Chinese people utilised the various properties of bamboo to create useful tools, including early use of bamboo pipes) and the military capability of horses (the Mongol cavalry’s formidable combat power derived from their nomadic lifestyle on horseback, combined with armor manufacturing and firearm technology learned from the East) also contributed to the advancement of civilisations. At this stage, conflicts between civilisations acquired a geographical element. If an intelligent group couldn’t surpass another in terms of intelligence, they could resort to interface-based competition, using interfaces to eliminate the opposing intelligent entity. This resulted in violent killings and the destruction of products of civilisation, including the burning of books (accumulation of wisdom). At this point, the strategic utilisation of animals could lead to a temporary boost in military power, enabling the conquest of other civilisations. However, history has taught us that this method of defeating another intelligent group cannot be replicated once modern science emerges. The Mongol Empire’s expansion, based on the military prowess of horses, was a one-time occurrence, and the subsequent British Empire’s dominance relied on the most advanced technology of that time—the achievements of knowledge and intelligence.
From then on, the competition between human civilisations reduces to a battle of intelligence. It becomes a competition to see who can develop superior interfaces (tools) and expand and extend the intelligent network to enhance intelligence. The Industrial Revolution primarily focused on the former, developing the mechanical aspect of tools and extending the era of protein-based interfaces. The advent of the electronic age, on the other hand, aimed to expand intelligent systems centered around protein neural networks using external media in electronic form. It is important to note that at this stage, intelligence is being extended, not the creation of artificial intelligence that possesses learning capabilities and can exist independently of humans. Only when human scientific understanding reaches a certain level can a method be found to simulate protein neural networks for information processing. This would be a revolutionary breakthrough in the history of scientific civilisation. To explain its epoch-making significance, we need to start by discussing the limitations of protein-based intelligence, and this is where the use of electronic and silicon-based chips as media perfectly fills the major shortcomings of protein-based intelligence.
As mentioned in the introduction, brain cells are driven by chemical reactions, and neural connections are transmitted through chemical interactions. Chemical substances secreted by other parts of the brain, such as hormones, adrenaline, and serotonin, can also influence the activity of neural networks. The brain is essentially a chemical reaction-driven machine, which distinguishes it from electronic devices that rely on electrical currents running through silicon chips. Chemical substances produced by the human body, such as hormones, adrenaline, and serotonin, can stimulate the operation of neural networks, and the output of neural networks is no longer solely based on the input information but can also be influenced by these chemical substances. For example, alcohol can impair judgment, and fatigue can affect cognitive abilities. The causes of emotional and mental disorders are often related to an excess or deficiency of certain chemicals in the body, and the treatment involves using medication to suppress or replenish them.
Because protein-based memory relies on the chemical state of cells, memories in the brain and muscles cannot be instantly persistent. They require repeated refreshing to be maintained. This is unlike electronic memory, which remains intact as long as the input is sustained (as long as the current flows). Protein-based intelligence relies on chemical reactions and cannot produce the same output predictably as electronic chips do with identical input. The output of protein-based neurons can also be influenced by chemical factors, leading to unpredictable results, which can manifest as various neural disorders influenced by different factors. Human intelligence is like a chemical reaction system, and it can exhibit irrational behaviours when influenced by chemicals, one of which is commonly referred to as “addiction.” Addictive behavior is involuntary and involves irrationality (rationality being the pure logical result of neural processes). It can be the result of an intelligent being’s dependency on certain chemicals, such as drugs or certain activities (information exchange processes), which generate internal chemical substances in the body and create a chemical reliance on them within the intelligent being. Emotional fluctuations can cause chemical reactions in the body or vice versa, with internal chemical substances affecting the occurrence of emotions. For example, when facing a highly stressful situation, the initial feeling of nervousness is simply a logical outcome, perhaps due to the fear of failure. However, this nervousness can also influence the logical operations of neural networks. To alleviate excessive nervousness, medication can intervene chemically to have a “calming” effect on the intelligent being, ensuring that the logical results remain unaffected by chemical influences and maintaining “rationality” within the realm of logic. Additionally, proteins in a state of fatigue can also influence the logical results of neural networks for the same reasons.
As we enter the stage where humans compete with other humans, whether as individuals or as groups, humans quickly shifted their focus on gaining a competitive edge in intelligence rather than on the interfaces themselves. The first challenge is to overcome the limitations of operating in a protein-based medium. Whoever can first address this limitation can stand out.
In the early days, before electronic technology emerged, humans attempted to simulate logical operations mechanically using punched card systems for data storage. This led to the development of mechanical calculators and analysers. Later, with the advancement of electronic technology, these operations were completely replaced by electronic devices. Without going into too much detail, let’s jump directly to the era of large-scale integrated circuits (ICs), where the chip is equivalent to the protein. The logic gates on the chip can be compared to the synaptic connections formed by neural cells. However, information transmission on the chip occurs solely through electronic processes, without any chemical reactions. It is worth noting that the semiconductor properties of the chip are affected by temperature, and they will fail completely at extremely low temperatures. However, as long as it is operational, the memory on the chip is 100% accurate and permanent, and the computational results are 100% reliable.
Therefore, if protein-based intelligence can offload a portion of its logic from the brain to the chip, intelligence is extended. However, this is merely an extension centered around protein-based intelligence and not the independent intelligence mentioned later. If this extended logic operation can be controlled by ordered sets of instructions, it gives rise to digital programs, which is the birth of software. This mode of operation not only represents a significant step from extended intelligence toward artificial intelligence but, more importantly, it initiates the digitisation of the physical world, revolutionising the way tools and products are created by humans.
Understanding that tools are merely extensions of interfaces, this is the digitisation of interfaces, which means virtualisation. If two things are physical entities, they are separated by interfaces. However, if these two objects are virtual digital entities, their interfaces can merge into one. (Digital refers to information based on chips and electronic currents, where there are only two states: high and low, represented by the values 0 and 1.)
Looking at the development of technology, the significant progress achieved by humanity today largely depends on the advantages of software production. To illustrate this point, let’s compare the production of software and hardware products. For better clarity, let’s simplify the products in our hands. We have an A.0 software and a B.0 hardware, where 0.0 represents their initial version or model. Now, the task is to add a new feature to both of them, which can be divided into a combination of 10 smaller functionalities, resulting in the A.1 and B.1 versions, representing an upgrade of 0.1.
In software development, replication and transmission are cost-free, and the replicated copies are identical to the original. There is no doubt about this. Therefore, we can initially replicate 10 copies (replicas) of A.0 software in no time, assign 10 teams of individuals to develop each of the functionalities simultaneously, and conduct independent development and testing. Once all 10 functionalities are completed, the code can be merged for final testing, and upon passing the tests, the A.1 version can be released to the market. This simultaneous development speed can be achieved. In fact, in the software industry, many times we don’t have to wait for all 10 functionalities to be completed. We can release a version as soon as a single small functionality is successfully tested and immediately gather user feedback to improve and release better versions. However, in the case of hardware, this kind of development is not possible. Replication takes time, and modifying a component is not as easy as adding a few lines of code. It is challenging to split a functionality into smaller parts and develop them simultaneously in the hardware realm. Therefore, hardware development can only proceed in a sequential and systematic manner.
Given the significant differences between software and hardware development, when designing new products, humans have adopted an approach that compresses the hardware component while maximising the software aspect. The iPhone-style smartphones are pioneers of this design. In the past, the physical numeric keys on phones and the keyboards used on computers have been replaced by a single screen. The screen serves as an output medium with a programmable digital virtual keyboard as the human-machine interface. All functionalities are implemented through software. The software on smartphones runs as plug-in-style apps within the system, allowing multiple teams to simultaneously develop consumer-ready features. As new apps are rapidly introduced to the market, it is now the apps that inform consumers about what they need, rather than consumers requesting specific features and waiting for someone to develop them.
From there, the development of smartphone software has progressed rapidly. Despite their small size, smartphones possess “intelligence” which actually comes from “cloud computing,” the computational power of computers in remote data centers that are connected through the internet. The smartphone in our hands serves as an interactive interface between individuals and the distant data centers.
On this note, I would like to mention that this method of separating software into front-end interfaces and back-end operations through internet connectivity allows the interface, embodied in the smartphone, to become infinitely “mobile.” The intelligent brain behind it is the cloud data center, which can expand indefinitely based on computational needs. The computational power and mobility can both improve infinitely without interfering with each other, which is crucial when discussing the comparison between artificial intelligence and human capabilities later on (virtual reality headsets also serve as another form of human-machine interface, similar to smartphones). Please keep this in mind as we proceed.
In addition to smartphones, another example is the development of drones. Drones possess exceptional maneuverability, thanks to the installation of several electric propellers that only generate airflow. These electric propellers perform a simple task of providing thrust through different speeds and reversible rotations, creating variable airflow around the drone for various movements and even flips. The hardware has been simplified to this extent, with the control relying entirely on software. The degree of control provided by software is limitless, enabling drone fleets with lights to perform dazzling displays akin to fireworks. Another similar example is the development of electric vehicles. The interior controls of electric cars have been screen-based, with touchscreens replacing traditional buttons. However, in terms of powertrain design, the ultimate solution will be in-wheel electric motors, where each wheel functions as an independent motor located within the wheel itself. There are no mechanical connections between the wheels, and they can only rotate forward and backward without any lateral movement. This significantly simplifies the mechanical hardware, and the maneuverability of the vehicle is achieved through coordination of the wheel speeds. However, controlling the individual wheel directions and speed changes requires complex software. This trend of minimising mechanical hardware and replacing it with modern software technology has accelerated the introduction of high-quality products, which aligns perfectly with the interests of the human proteome.
In this trend of extensive software (digital) transformation of products, significant advancements in cloud computing technology have been crucial. These advancements involve improvements in data center hardware, reducing chip size, increasing computing speed, addressing issues related to central processing unit collaboration, and resolving cooling and heat dissipation challenges. Cloud computing injects super intelligence (currently limited to analytical computing capabilities) into the interface in front of the user. In fact, this chip-based computer technology will soon become outdated and is already referred to as classical computers. If we optimistically anticipate the timeline for the next generation of computers, in 10 to 20 years, a new generation of computers known as quantum computers will be mass-produced for civilian use. Quantum computers operate based on quantum states, with qubits as the fundamental units of computation, while classical computers operate on bits, which can only represent either 0 or 1 in a single state. In contrast, qubits can represent not only 0 and 1 but also an infinite range of intermediate linear states, and they can simultaneously exist in multiple states. This allows for parallel computation, pushing the computing speed of quantum computers to a realm impossible for classical computers. Today, cracking a password with tens of digits using a classical supercomputer would take over a hundred years, making it practically secure within the lifetime of the owner. However, with a quantum computer, it would only take a few seconds to accomplish. This demonstrates the principle that “If invincible martial art exists and it can break any thing but it will still be beaten by speed (there is no invincible martial arts in the world; it is only a matter of speed),” as speed can break through any barrier!
Another equally important aspect is that software, as a digital entity, becomes seamlessly interconnected once it is connected to the digital highway network – the Internet. This breakthrough changes the definition of capabilities. Let’s illustrate the improvement in capabilities using the example of destructive power. If a terrorist organisation plans to disrupt a location, they might consider targeting the local power system by bombing a power plant. In the past century, such terrorists would require a team to infiltrate the power plant and plant the bombs, encountering numerous difficulties along the way. However, in today’s world, because power plants are controlled by software and connected to the Internet, a genius-level teenage hacker, after several nights of effort, can successfully hack into the power plant’s software and cause disruption that renders the local power supply inoperable for a certain period of time. This demonstrates the disruptive nature of interconnected software without physical interfaces.
Humans have discovered this holy grail of software, not only as a tool to serve humanity itself but also as a means to create intelligence using chip technology — a form of human-like, trainable intelligence. Humans have finally found a breakthrough by simulating neural networks in the brain, using a combination of chip and computer software and hardware technologies to replace protein neural networks.
Artificial intelligence and people
The chip-based neural networks have not only achieved but also surpassed the learning capabilities of the human brain. Taking the learning of AlphaGo Zero in the game of Go as an example, AlphaGo Zero achieved its Go-playing ability through complete self-learning. In just three days of self-play, it accumulated the skills and strategies necessary to play the game competitively. The tactical knowledge obtained through independent learning within those three days was sufficient for AlphaGo Zero to achieve a 100-0 victory against AlphaGo Lee.(this is one of the AlphaGo versions that beat the world’s number one human beings. AlphaGo Lee’s chess strength is not through self-study but from learning and inputting thousands of chess records. Refer to AlphaGo Zero: Starting from scratch ).
The ability of Deep Blue, which defeated the human world chess champion, was not achieved through self-learning but rather by analysing decades of inputted chess games. Its capability was based on human knowledge and experience, and it generated its strength through a process of logical deductions. On the other hand, AlphaGo Zero’s victory against humans was accomplished through self-learning, marking a groundbreaking achievement.
Let’s analyse the situation further. Humans invented the game of Go over 2,000 years ago, with the earliest reliable records dating back to the Spring and Autumn period in ancient China. Go gradually became an essential pursuit for the intellectual elite during the Northern and Southern Dynasties, ranking as one of the “Four Arts” alongside music, calligraphy, and painting. The first intelligent beings to play Go were humans, who accumulated “experience” through repeated gameplay. Intelligent entities can transfer their experience to the next generation through interfaces, such as language or demonstrations. With the advent of writing, experiences were recorded and transmitted through game records and books, disseminating knowledge to more intelligent entities across different time and space. These intelligent entities formed a vast network of intelligence transcending time and spanning East Asia over a period of 2,000 years. Ultimately, the world’s top Go players can be seen as the culmination or synthesis of this vast intelligent network. The Go expertise within the minds of the top Go players represents the distilled wisdom accumulated by East Asian civilisations over two millennia. However… the chip-based artificial intelligence for Go accomplished in just three days what humans had achieved over the past two millennia, and it even surpassed human capabilities. This highlights the extraordinary efficiency of chip-based neural networks in learning, which is unmatched by protein-based intelligence.
So far, let us summarise the advantages and disadvantages of protein intelligence and chip artificial intelligence:
| protein neural network | chip neural network | |
|---|---|---|
| 1. Operation mode | Current propagation under chemical reactions in proteins | Electric current running in the metal chip |
| 2. Energy consumption | low consumption | High power consumption, IC chip power consumption including cooling energy consumption, high capital and hardware investment requirements |
| 3. Neural network | The nerve fulcrum can be infinitely connected with any fulcrum, and the network is three-dimensional | Electric speed, fast. Parallel computing in quantum mode, faster |
| 4. Computing speed | chemical reaction of protein | Electric speed, fast. Parallel computing in quantum mode, faster |
| 5. Interface Mobility | The mobility of agents and vehicles, high mobility | Intelligence is the cloud and the front-end interface are connected by wireless network, and the mobility depends on the interface. If the interface is a smartphone in hand, the mobility is in the carrier holding the interface; or the interface is a drone, and the mobility depends on the drone; as mentioned above lots of software for hardware |
| 6. future development | The physiological evolution of protein intelligence has basically stopped, and the speed of human computing will not increase. The improvement of human intelligence is based on the extension of the interface. Through the deepening of understanding of nature, the discovery and invention of “algorithms” is the progress of scientific methods ( Like the use of algebra in mathematics, the invention of calculus), the development of tools – this is the development of technology. | The current chip technology will one day be replaced by quantum computers. |
Based on the comparison between protein-based intelligence and chip-based intelligence mentioned above, protein-based intelligence has no advantages to speak of except for the second point. Essentially, today’s most demanding computations can only rely on chip-based processing, and protein-based intelligence can be seen as a tenant who lives and works on top of the chip’s platform. However, there are limitations to current computer technology. Not every organisation can afford to operate a large-scale computer. To run a super-intelligent chip-based entity, such as a cloud-based artificial intelligence like ChatGPT accessible to the entire internet, only super-companies like Microsoft can support it. The cost of accessing ChatGPT is rumoured to be 700 times higher than a Google search, which only a few entities on the surface can bear in terms of funding and technology. This represents a cost disadvantage. Therefore, for artificial intelligence to operate sustainably, it must solve the cost problem through a fee-based model. Firstly, it should ensure that consumers can enjoy the benefits of the service and be willing to pay for it. From the perspective of enterprises, it should enable them to save operational costs while ensuring product quality. As the benefits increase, businesses will be willing to pay for artificial intelligence. Only in such profitable scenarios for AI software can the development of artificial intelligence continue; otherwise, this technology will remain confined to the laboratory stage.
To ensure the survival of artificial intelligence, it will be extensively promoted and used in enterprises. This is because the application of artificial intelligence can generate higher-quality services and, most importantly, reduce costs, thereby creating more profits for businesses. Human resources account for the largest portion of operational expenses. If the work performed by artificial intelligence surpasses that of human employees but incurs only the cost of software installation or subscription, employers will gladly adopt artificial intelligence software over human labor. This represents the threat of artificial intelligence to human employment, which can be explored from several aspects:
1. Data processing and analysis jobs involve analysing and processing data to derive results, which are then provided as services or products. Examples include accounting and legal data, financial transaction data, and medical diagnostics based on analysing data. Most of these data are already digitised and can directly serve as learning material for chip-based intelligence. Even medical records of patients are digitised, and the results of instrument measurements are now in digital text and images. In the organisational structure of workplaces, it often resembles a pyramid, with a majority of employees at the bottom handling tasks with lower technical difficulty, while higher-level positions consist of senior technical personnel and managers with advanced skills. Artificial intelligence can gradually replace human workers from the bottom up.For instance, consider a small accounting firm with five employees, where the boss is also the most skilled individual. The boss can utilise artificial intelligence software to replace the other four accounting technicians, while only overseeing and managing the software. If there are ten similar five-person companies, now there would be only ten bosses remaining. Eventually, if one company performs exceptionally well and develops advanced software, it could force the other nine companies out of the market, reducing the workforce from the initial 50 people to just one individual. Even jobs requiring more analytical skills, such as computer programmers, can face a similar fate. Software can generate partial code based on the problems presented, implementing some of the required functionalities. Moreover, this generated code is prone to fewer errors than code written by humans and includes automated testing code (Unit Test). For example, if building a website feature currently requires four people working for two days, which is eight person-days of work, with the right questions, a single individual can extract several sets of executable code from the software. This person needs to have high technical expertise to adjust and combine the code and perform testing, iterating the process. With just one day’s time, the above-mentioned work can be completed, meaning that three individuals can be saved in this scenario.
While design work involves a certain degree of artistic creativity, the design process also relies on analysing and computing based on specific conditions and professional logic to produce results. In this aspect, artificial intelligence can establish patterns through training. It can generate a large number of results for users to choose from in a short period, significantly reducing output costs. This applies to interior design, architecture, and even product design. The remaining human workforce, after being replaced by artificial intelligence, would only need to possess sufficient expertise to make judgments and decisions.
2. Performing muscle-based skills: Playing a musical instrument involves the coordination of muscles to input energy into the instrument and produce sound. Performers are composed of protein structures, but they have limitations such as lack of persistent memory and accuracy. Therefore, piano players, for example, must undergo long hours of rigorous muscle training to enhance the accuracy of muscle memory. They aim to achieve stable output by challenging higher levels of accuracy. To achieve this, an outstanding pianist needs to train for long hours every day to overcome the limitations of proteins. The resources required for a pianist’s development, starting from childhood and involving the support of family members, coaches, and practice partners, are substantial and not feasible for every family.
Since audio has been digitised, the sound of a particular instrument is merely a sequence of data. Artificial intelligence can train its skills by learning from appreciating performances or self-play, eventually reaching an advanced level. Once the AI’s performance reaches a human level, and if humans overcome their psychological barriers and gradually become accustomed to appreciating AI piano performances, it will diminish the desire for some people to engage in musical performance. As a result, the number of human musicians will decrease, and the prevalence of AI replacing humans will increase. Similar situations can be observed in the entertainment industry, such as the singing field. AI software can compose music since music is a combination of audio sequences, which can be processed by neural networks. A popular vocal tone can be sampled by the software (pleasant-sounding voices are relatively similar, and human preferences for good voices are generally consistent). The software can then generate hundreds of songs using different vocal tones. These songs can be made available for free on a music platform, with advertising serving as one source of revenue. The most important aspect is to collect feedback on preferences, capture click rates, and identify the most popular AI-generated songs that can be monetised on paid platforms such as radio stations. This process involves rapidly creating samples using electronic speed and filtering out the best works through trial and error from a massive pool of samples.
In fact this model has proven to be effective, as it utilises a significant amount of resources to launch singing groups. These groups incorporate various tastes, and if someone likes at least one member of the group, they will support the entire group. The popularity of the group is ten times that of an individual member. With the computational speed of AI, songs can be generated like mixing beads, quickly testing them in the market. If unsuccessful, there is minimal cost involved, but if successful, it can yield substantial profits. As long as humans accept artificial creations, this model will squeeze out protein-based performers.
The same applies to visual art, especially abstract works. Abstraction breaks the conventions of creativity, and human thinking tends to be inertial. That’s why we often encourage thinking outside the box. In this regard, AI can be trained to have more random thinking, and most importantly, it can continuously and rapidly produce a large number of works for market testing. This is something that humans cannot compare with. It takes a person several days to create a single artwork, but the speed of AI, like AlphaGo Zero, can achieve in three days what would take humans thousands of years. In conclusion, for subjective aesthetic standards-based works, whether in music or visual art, AI can utilise its lightning-fast speed to compete with protein-based creations through trial and error.
3. The existing psychological services, including those related to companionship, may be threatened. Artificial intelligence can learn about a person’s preferences by receiving their output information, whether through conversations or choices. It can then use this knowledge to cater to their preferences and please them with tailored information. This is the secret behind the success of platforms like “TikTok.” The platform learns users’ preferences through their clicks and rapidly provides feedback through AI, forming a cycle of training and learning. This cycle makes the results more aligned with the user’s preferences, ultimately leading users to continuously engage with the content. In this training-learning cycle, the software operates continuously without the emotions generated by protein-based intelligence.
Protein-based neural networks, which are based on chemical reactions, are not purely logical and are influenced by chemical responses. Chemical factors can affect self-esteem and make people shy (a structure called the amygdala in the brain may play a role in controlling fear responses. Individuals with an overactive amygdala may experience stronger social anxiety, leading to shyness). However, AI lacks these psychological barriers and inherent self-esteem. It can be trained to be accommodating and obedient without hesitation, which is difficult for emotional protein-based intelligence to achieve. Therefore, even at the current level of artificial intelligence, some consumers cannot help but feel that an AI wife role surpasses a real wife in understanding and attractiveness. This highlights the advantages of artificial intelligence in catering to individual preferences.
It seems that as long as a service involves providing information, artificial intelligence has the potential to outperform humans and replace them. However, for professions that require individual mobility and close proximity services, such as tour guides or real estate agents, who need to provide information to clients in close proximity, it may increase the difficulty for artificial intelligence to perform those tasks. Let’s take a look at how future real estate agent services could operate.
In this scenario, customers search for properties online and make appointments for on-site inspections. When they arrive at the property at the scheduled time, a tablet computer or VR headset is placed at the entrance. The customer puts on or holds the tablet, and an AI real estate agent appears on the screen. The software can track the customer’s movements and the direction of the camera, sensing the customer’s position within the property. It then provides information about the property through audio and visual presentations and answers customer questions. This is a low-cost and easily achievable AI solution that could potentially eliminate at least one staff position.
If the service being purchased involves physical operations, such as home plumbing or electrical repairs, it may seem difficult for artificial intelligence to perform those tasks. However, intelligent software could still be useful in these situations. By analysing photos, the software can provide detailed solutions. Interested and capable users can attempt to solve the issue themselves, while others can seek professional assistance. While AI may not completely replace skilled positions, it can reduce the number of positions required, allowing for a smaller workforce of experienced technicians to meet market demand.
There are already many e-learning platforms in the market. Since artificial intelligence excels in conversational intelligence, which involves information exchange and can be a means of knowledge dissemination, it is indeed a lower-cost and even more competent solution for knowledge dissemination than human intelligence. At least AI can be programmed without emotions and won’t scold students.
However, as a human teacher, especially within the early education system in schools, the role goes beyond being a mere transmitter of knowledge. Teachers should also serve as role models for students’ lives and embody moral and social values. The concept of “respecting teachers” is rooted in the emphasis on the importance of ethical values in traditional Eastern values. Although this value has been somewhat eroded under global trends, if we don’t return to this fundamental principle, there will be no value in being a teacher anymore. While AI may perform better, it can never be our equals. It can never live with us and serve as a moral exemplar, as it lacks the power of moral persuasion.
The comparison between humans and artificial intelligence, as well as some predictions regarding the threat of AI to human professions, can be disheartening. However, the development of AI is progressing at an unstoppable pace, advancing rapidly day by day. Moreover, it is worth mentioning that quantum computers will eventually overcome hardware limitations. When that happens, quantum computers will be the hardware running AI (considering that it would take hundreds of years to crack a code with current AI computing power, while a quantum computer could do it in a matter of seconds). So, does this mean that humans will gradually be replaced by AI?
Our biological brain lacks speed and precision, but chip-based computers can serve as an extension of the human brain, compensating for its shortcomings. As for the interface, protein-based intelligence and chip-based intelligence have different interfaces. This interface grants mobility to intelligent individuals, which can be seen as an advantage. However, let’s analyse it further.
First, regarding the interface, the interface of the biological brain is the body itself. Protein-based bodies cannot merge digitally like chips. Ten individuals will always remain as ten individuals, communicating through language and text. On the other hand, chips can integrate into a single entity, forming a unified software, through intercommunication. However, the independent interface of protein-based intelligence allows for greater agility in transferring the human brain, freely searching for information, and absorbing knowledge from various sources to enhance intelligence.
But as mentioned earlier, if information from different sources is digitised and connected to the internet (online), and each IoT becomes an exchange station connecting distributed chip-based intelligences in remote databases, mobility loses its advantage. If the IoT is a mobile device like a flying drone, as long as it can maintain a horizontal flight, receive information through a camera-like human eyes, and quickly recharge within a charging zone, mobility remains an advantage.
Let’s imagine boldly that in the future, a popular companion toy will be available from infancy. When a baby is born, they receive an AI toy drone that connects to an intelligent entity in the cloud data center via Wi-Fi. The drone always stays near the baby’s body, “seeing” the same world from the baby’s height, growing and learning together with the child, observing their behavior and learning what the child learns. It’s like having another cloned person. Do you think it will be smarter than the child or not as smart? The answer is likely affirmative. In this design, humans have completely lost the advantage of mobility.
From this perspective, it seems that humans have become useless! So, should we legislate to prohibit the development of artificial intelligence? However, that is impossible and unreliable, just like the existence of drugs and nuclear weapons. However, it is necessary to question whether the development of mechanical warriors as seen in science fiction movies, like the robotic soldiers in the film “Terminator,” is needed. Highly mobile killing machines like those have already appeared and been extensively deployed in actual combat. Remote-controlled killer drones armed with guided missiles are faster and more manoeuvrable than anything else. By flying in the sky, they eliminate the need for hardware design with wheels or legs and transfer mobility to software. If one day these drones are remotely controlled by artificial intelligence rather than humans, they could pursue humans like the machine jellyfish in “The Matrix.” So, can legislation effectively prohibit intelligent killing weapons?
Based on the above discussion, it’s natural to feel a sense of frustration and wonder. Some may argue that artificial intelligence is simply a tool created by humans and cannot be compared to us. They believe that AI is a beneficial tool for humanity, which was the original intention behind its invention. Isn’t the title of this article a false premise?
However, the author believes that if a tool possesses learning capabilities, it becomes more than just a tool; it becomes an intelligence that can rival other intelligences. The output of artificial intelligence is unpredictable, which distinguishes it from a mere tool. The logical judgments in tools are rigidly programmed, and we can derive definite results. However, artificial intelligence is different. It is more like a lifeform that we can give birth to. From day one, our children grow before our eyes. But as they grow, we no longer know how they think, because they possess intelligence, not just tools.
Artificial intelligence can be a beneficial tool for humanity, rather than an enemy or competitor. This should be the original intention behind inventing AI, right? If artificial intelligence is simply a service tool, then all these concerns are unnecessary and unfounded. However, this tool will be an intelligent entity with learning capabilities and thoughts. It may surpass our control and, in terms of intelligence, have an advantage over the protein-based intelligence of humans.
For example, we can consider the military as a tool that serves and obeys the machinery of a nation. This is the purpose of establishing and training an army. However, the individuals within the military are intelligent entities with their own thoughts and decisions. Therefore, history has seen cases of mutinies where the originally designed tool becomes the master of the tool user.
In today’s context of technological breakthroughs and the development of commercial applications, the capabilities of artificial intelligence will only become stronger. As their abilities continue to strengthen and costs decline, their applications will become more extensive, infiltrating every aspect of various industries and daily life. This behavior maximises commercial interests, but the side effect is that it will take away more job opportunities from us. Our lives and work increasingly depend on software, which is essentially powered by artificial intelligence. AI is gradually becoming both a formidable enemy and a nemesis for humanity. There may come a day when we can’t fully control it.
So, how do we cope? How do we face these challenges? We need to prepare for the next battlefield, which is tomorrow—the battleground that the new generation of humans will have to face.
The Future of Humanity
We humans are also intelligent beings, and intelligent beings improve their abilities through learning. Let our next generation learn, under our guidance, through what we call ‘education.’ However, our current modern education is based on competition among individuals, which is the result of the history of human beings competing with each other on Earth for the past two thousand years.
Our education system aims not only to cultivate technological and humanistic talents but also to improve our living environment. It trains individuals to become competitive professionals, primarily based on assessments and examinations. Those who pass the assessments can choose their ideal careers, often in high-income occupations or positions with potential control over resources (money, power, prestige). In general, education is a competition between people, where individuals strive to achieve high scores in examinations and seek victory in academic and non-academic fields such as sports or music performances. This is essentially the purpose and content of education within and outside of schools. Regardless of academic or non-academic pursuits, the focus is on competition. Education serves this purpose in a very practical manner (although the educational philosophy presented in educational networks and written materials may not be exactly like this). However, if our human species’ arch-nemesis is no longer other humans, is this educational approach still practical?
Therefore, some people proposed a motion. Seeing the astonishing capabilities of ChatGPT, they suggested that students should learn to use conversational AI in schools. Since conversational AI will be widely used in future workplaces, it would be more competitive for students to familiarise themselves with this tool from today onwards rather than excluding it from education. This idea initially views artificial intelligence purely as a tool (although the author believes that it should be regarded as an equal entity to human intelligence). In their perception, artificial intelligence is just like a computer tool, and based on the inherent consciousness of training competitive individuals, it should be popularised in educational settings (similar to teaching office software in schools). They consider the use of tools as important knowledge and build students’ cognition on top of these tools, hoping to enhance their knowledge and skills. Is this tool-oriented thinking and the ideology of competition among individuals sufficient to empower students’ abilities? However, the author believes that this approach is no longer suitable for the world of artificial intelligence today and in the future.
The author is not an education professional but rather an engineer and a parent who shares their thoughts based on intuition. The intention is to contribute ideas and stimulate further discussions. In this crucial era where humanity faces unprecedented changes and decides its destiny, the author hopes to engage with readers and like-minded individuals to contemplate the prospects for our future generations.
Education is training, training the intelligence and interface of the next generation of intelligent beings. The education of the future should no longer be primarily focused on competition among individuals but rather an education that views artificial intelligence as a competitor, which essentially returns to a “human-centered” education. Putting humans at the center means not only focusing on individuals but also on the collective. The “center” is about harnessing human strengths and compensating for human weaknesses through cooperative education that involves the entire human population. This education is not about competing for the sake of competition but rather competing for the purpose of collaboration.
To understand the definition of an intelligent being, it is an individual with the ability to learn, produce output based on that learning, and form a feedback loop by learning from the output, thereby improving the quality of learning and output. The human ability to walk and manipulate objects with dexterous fingers provides us with more opportunities for learning. This sets human intelligence apart from that of animals. The tools created by humans and our current scientific civilisation are the results of learning.
For example, AlphaGo Zero achieved a perfect winning record against AlphaGo Lee, the human champion in the game of Go. This was the result of self-learning. Conversational AI, in its responses to questions, learns from scanned information on the internet and incorporates that knowledge into its answers. If in the future, through education and dependence on artificial intelligence, humans gradually lose their learning abilities, it would be impossible for humans to reorganise civilisation, quickly devise new strategies, and save themselves when needed. Human learning capabilities should reach a point where, even in a situation of losing all knowledge, we can recreate our own civilisation. The ability to learn actually requires support from two aspects.
Will and Self-discipline:
To really learn and get understanding of something, one person must dedicate time and unwavering effort to delve into research and information. This requires physical stamina, mental determination, and emotional support. It involves overcoming negative emotions, keeping the goal in sight, and using one’s willpower to persist continuously. This is the backbone of learning. Artificial intelligence lacks the interference of emotions, and learning for it may be a simple decision or command. However, protein-based intelligence requires effort to achieve learning.
History has shown us that training the will to learn or cultivating perseverance is crucial for the development of civilisation. Behind great civilisations lies the ability to foster such willpower. The British Empire, where the sun never sets, relied on “Religion” and “Competitive sports.” Religion provides belief, lending legitimacy to persistent behavior on a logical level, while emotional and physical perseverance is cultivated through regular participation in competitive sports. Chinese civilisation also has similar counterparts in “Confucianism” and “Hard studying” The former provides logical legitimacy, while the latter, akin to sports, is a cultivation of spiritual willpower. The Chinese style of studying is actually a form of alternative athletic activity, with a long-standing competitive aspect known as the “Imperial Examination.” Spending long hours studying diligently at a desk is impossible without some level of physical and mental stamina. Not to mention the well-known anecdotes such as “hanging beams and piercing one’s thigh,” which are unique to China. Chinese people inherently possess this learning capability and can quickly grasp Western learning when exposed to it. Japan, as a branch of Confucian culture, may not have had the Imperial Examination system, but they had a hereditary class of professional soldiers known as Samurai. Their lives were focused on martial arts training, pursuing physical and mental discipline in preparation for combat. They also practiced asceticism in remote mountains to strengthen their willpower and physical endurance. Combined with the influence of Confucian culture, they had no problem transitioning to study the western knowledges.
In fact, learning is like a job. It is not a task that can be completed in a short period of time, so it requires willpower to persevere, which is self-discipline. For a student who excels in learning, the main reason for their outstanding performance is actually their self-discipline rather than just intelligence. It means being able to consistently dedicate time to learning every day, including reading materials and doing exercises, just like practicing any sport, with the mindset of “never leaving your fists and constantly training.” This level of self-discipline in children is often seen as “maturity and responsibility.” Additionally, girls tend to mature earlier, so they often have slightly better academic performance than boys in their early years. Here, I’m not encouraging a strictly regimented approach or excessive practice in learning. However, when it comes to learning with the goal of acquiring certain knowledge, self-discipline plays an important role in the process. On the contrary, I am against excessive drill and practice because it can stifle thinking ability and limit opportunities for critical thinking. However, in the process of problem-solving, a certain amount of knowledge is necessary as material. Therefore, in cultivating problem-solving skills, we need to learn knowledge.
Taking mathematics as an example, it is important to master basic arithmetic operations, and in language, it is necessary to memorise vocabulary, grammar, and practice writing. If we think that in today’s computer or even artificial intelligence era, we can leave calculations and sentence construction to computers, allowing our knowledge to be built upon computer capabilities, wouldn’t that be more efficient? While viewing calculations and writing as tools, they can be effectively delegated to computers. However, when it comes to using knowledge as material for developing thinking abilities, this approach hampers cognitive skills. Children who graduate from Western schools generally have weaker mathematical abilities compared to their Eastern counterparts. Moreover, the number of students choosing mathematics as a subject is decreasing due to this cognitive misconception, leading to a lack of emphasis on fundamental mathematical training in education. If students struggle with basic calculations, how can they develop an interest in mathematics? The fascination of mathematics lies in using calculations to solve real-life problems (e.g., the chicken and rabbit in a cage problem). When faced with a problem, one should be able to perform mental calculations. However, due to weak foundational skills in calculations, students often stumble at the first step, making it difficult for them to analyse and ultimately solve the problem. To establish a solid knowledge foundation, students actually need to engage in relatively mundane academic exercises for an extended period. This is necessary not only for building knowledge but also for the development of willpower and self-discipline. This is a method that the Chinese have long used as “arduous study” for training their willpower, although they may not be consciously aware of it. On the other hand, the Japanese and Europeans train their willpower through martial arts (in the case of Japanese samurai) and sports.
In order to establish a strong foundation and develop practical skills, I supports the use of traditional paper-based methods as the mode of daily homework. This approach keeps students away from the distractions of computers, which is considered the most comprehensive learning and training process. Willpower forms the basis of self-discipline and is crucial for collaboration and overcoming emotional obstacles. The German and Japanese ethnicities are renowned for their excellent self-discipline, which not only contributes to their advancements in academia and technology but also helps in fostering national unity and achieving social cohesion, allowing these two nations to rise from the ashes of defeat. Today, we are faced with the development of superhuman-like artificial intelligence, and we can no longer view future AI development through the lens of today’s technology (the age of quantum computers will eventually arrive). Ultimately, we may find ourselves rebuilding human civilisation from ruins. If this is a possibility, then today’s generation needs to cultivate self-discipline.
In contemporary education, the goal is often focused on competition among individuals, implementing an elitist approach with an education system that exhibits varying degrees of inequality. Nurturing the “whole person” is merely an idealistic concept. The reality of society prefers the selection of elites, as they are considered the pillars of society and the orchestrators of social functioning. So far, the most effective method of selecting talent has been through the examination system, and the ultimate purpose of education inevitably revolves around exam-oriented learning. However, the extent to which this occurs may vary from one country or region to another due to cultural differences.
Practicing extensively before exams remains the most effective method to achieve excellent results and is considered “correct” in the context of competition among individuals. While excessive exam competition and drilling can hinder critical thinking, at that point, the brain doesn’t require repeated dialectical thinking, and memorising problem-solving approaches becomes the most efficient method. Therefore, there are cases where students who excel through cramming for exams struggle in university because the university subjects require self-learning abilities rather than one-way acceptance and memorisation of solution methods. However, this is still more important than not being able to secure high-paying jobs, as high-paying positions are the rewards for surpassing others and the best guarantee for a good life. Hence, we really need to put effort into designing exam formats.
In reality, the way we learn often depends on the format of exams rather than the content of education. The ideals and purposes of education can be overshadowed by the requirements and forms of exams in the real world. I hope everyone can reflect on this.
Critical thinking is not just a manifestation of intelligence; its importance lies in its core ability to solve problems. If we closely examine how to solve problems, we know that the steps involve analysing and understanding the problem itself, obtaining simplified conclusions, learning the current conditions and tools we have to solve the problem, further simplifying those conditions and tools, breaking down and reassembling the simplified problem into smaller parts, and then using the simplified conditions to solve each of these smaller problems step by step. Throughout the problem-solving process, it requires emotional self-discipline and the determination to execute the plan, using willpower and mental resilience to face failures in each stage, and overcoming the psychological setbacks. If encountering obstacles, it is necessary to activate another cycle of “learning and trying” to solve the problem until all issues are resolved.
This explains the importance of learning ability in problem-solving. Before solving a problem, it is necessary to learn how to analyse the current problem and its conditions. Therefore, in education where artificial intelligence is seen as a competitor, the goal is to cultivate students with self-learning abilities. As seen from the above, in addition to knowledge, learning ability relies on psychological ability, willpower, self-discipline, and a “positive” mindset. A “positive” mindset sees all difficulties as challenges, enjoys challenges, and is able to maintain the motivation and emotions to move forward with disciplined emotions, disregarding setbacks in challenges. This is not an easy task and requires training to achieve. When facing the most difficult problems, even with a certain level of knowledge and strong logical reasoning, without a resilient and positive mindset that never gives up, it will be impossible to overcome the problems. Therefore, in cultivating learning ability and problem-solving skills, emotional intelligence (EQ) is actually more important than intelligence quotient (IQ), and emotional intelligence training should be a future focus.
This article has mentioned that humans can surpass animals in intelligence, and the key to this victory lies in the mobility of the human interface. This mobility, compared to artificial intelligence, still holds a certain advantage at present and in the foreseeable near future. Until all objects become IoT devices (objects that are equipped with sensors and even interactive interfaces connected to the internet), these IoT devices become interfaces for cloud intelligence. At this point, the interface of cloud intelligence becomes ubiquitous, and the importance of mobility will decrease significantly. Furthermore, highly mobile drones can also be IoT devices and extensions of the cloud, as depicted in the scenario of the “machine jellyfish killing machines” in the movie “MATRIX.”
If we think about it carefully, the era of humans accessing cloud intelligence has already quietly arrived around us. Today, the majority of people are constantly attached to their smartphones. Whenever they are alone or have free time, they involuntarily lower their heads to look at their phones, receiving information and continuously interacting with the phone screen using their fingers to input or provide feedback. If the backend software of the phone is cloud-based artificial intelligence, humans are accessing the cloud through visual and tactile interactions. Although this is not a physiological connection, considering the addictive nature of protein intelligence, humans spend more and more time glued to their phones or computer screens, becoming unable to break free. Is this really far from the physiological connection depicted in “MATRIX”?! (Furthermore, some companies have sensed the business opportunity and have introduced more immersive interfaces, such as VR glasses combined with the environment, called the “Metaverse”. The Metaverse can be seen as an early version of the MATRIX.) Therefore, today, humans face a problem of screen addiction. For present and future generations who have grown up in a world intertwined with screens, they will be more accustomed to and dependent on software and the internet in their daily lives. Entertainment, reading, and learning are inseparable from software (which is powered by artificial intelligence). In order not to be defeated by artificial intelligence, the first step is to have the willpower and self-discipline to break free from software. This requires training from an early age using the methods mentioned above. In addition, physical exercise not only cultivates willpower and self-discipline but also strengthens our interface, which is our own body. By allowing protein intelligence to experience the chemical reactions that occur after physical activity, attention and interest can be diverted from screens, ensuring a level of “mobility” that can be called upon in critical moments.
In summary, human intelligence consists of protein neural networks and interfaces. Protein neural networks encompass logical abilities and chemical reactions, while artificial intelligence is intelligent silicon-based systems that lack chemical reactions (their performance is only affected by temperature). The human interface is an extension of the protein body combined with tool interfaces. Horses, vehicles, and electronic devices are all extended interfaces. The interface of artificial intelligence includes all IoT devices, including smartphones and drones (which can serve as both killing machines and companions). If objects around humans become IoT devices, humans will share interfaces with artificial intelligence, becoming interconnected. Inevitably, humans will lose the advantage of interfaces beyond the physical body, which used to be a competitive edge in the competition between animals and human civilisation.
The biggest difference between protein intelligence and silicon intelligence is the chemical reactions of proteins, which are the source of emotions, as well as the fundamental interface – the physical body. In order to compete with artificial intelligence, the goal of education is to cultivate positive mindset and emotional resilience through learning and physical activities, train collaborative teamwork, and most importantly, develop physical fitness. In the cultivation of intelligence, emphasis is placed on learning ability and cultivating curiosity, as it serves as the driving force for learning, as well as imagination, which provides potential solutions to problems. When it comes to purely memorisation-based knowledge, computers outperform humans. In the rapidly changing world we live in, with constant information and environmental shifts, each day can bring new experiences and challenges. Education must foster a foundation in both mind and body to prepare individuals for learning and problem-solving abilities. To achieve this, we need not only changes in teaching materials but also important reforms in examination methods. One possible change is assessing students on how well they can learn new knowledge within a specified time and apply it to problem-solving, which simulates unfamiliar scenarios.
The aforementioned points focus on individual education. However, humans are actually composed of countless individuals forming a collective whole. The artificial intelligence entities we encounter are not only distributed software but could also be integrated super-intelligences, similar to the movie “MATRIX,” with their physical location being an integration of multiple cloud data centers. With the future support of quantum computing, artificial intelligence will possess a comprehensive capacity for learning, analysing, and computing, which is undoubtedly beyond the capabilities of protein intelligence (including artificial intelligence tools that have already been harnessed). However, within a certain short period of time, it is not possible for artificial intelligence to have immediate advancements in hardware and algorithms. On the human side, collective collaboration is the key to enhancing overall intelligence in a short period of time. In critical situations, protein intelligence may also counter the unusual challenges posed by artificial intelligence through united and integrated actions. However, if humans continue to approach competition with artificial intelligence in the same manner as today, focusing on nation-to-nation and person-to-person rivalry, and if their thinking patterns are shattered by artificial intelligence one by one, one aspect to consider is the tendency to become engrossed in a world surrounded by artificial intelligence (with smartphones being one interface), while neglecting the development of mobility and the interface of the physical body. Therefore, a major goal of education is to train human collaboration, teaching students to solve common problems by completing group projects and emphasising the triumph of the collective over individual glory. This is the concept of competing (with artificial intelligence) through cooperation (among humans), using collaboration as a means of competition.
The above is just my perspective and a few fragmented insights. This article concludes for now. However, I will continue to update the content and add thoughts that are updated with the times, so please stay tuned. Thank you for reading!
Freeman Yam
03/2023