In this blog post, we explore whether artificial intelligence can perform human emotions and intuition, and even the role of the right brain, based on the principles of emotion formation and the functioning of artificial synapses.
The human brain consists of the right and left hemispheres. Most higher mental functions occur through the interaction of both hemispheres, but the right and left hemispheres are primarily responsible for different functions. The left hemisphere is responsible for linguistic, mathematical, analytical, logical, and rational functions, while the right hemisphere is responsible for non-linguistic, spatial, intuitive, and emotional functions. Brian Christian, author of “The Most Human Human” (2012), argues that the right brain is one of the key differences between humans and artificial intelligence. While the left brain’s functions of calculation and logic can be adequately replicated by current computers, the emotional aspects handled by the right brain are difficult for artificial intelligence to perform. However, I agree with Brian’s opposing view that if we can figure out how emotions, moods, and intuition are formed within the brain, we can apply that knowledge to artificial intelligence. I will examine emotions, which are the most representative functions of the right brain and are most commonly claimed to be impossible for artificial intelligence to possess, and discuss the possibility of artificial intelligence mimicking the right brain’s domain.
First, the occurrence and expression of emotions in response to situations are similar to artificial intelligence algorithms. Emotions are formed based on evolution and memory. Mammals have more developed limbic systems and cerebral cortexes than other organisms. The interaction between these two developed parts enables complex responses, which is what we call emotions. As mammals evolved with the goals of survival and reproduction, they developed common emotions such as love, fear, anger, and sadness through mating for reproduction and fleeing from predators for survival. This can be evidenced by the fact that newborn infants who have not received any education laugh when they are happy and cry when they are unhappy. Emotions that are possessed without learning are called primary emotions, and they form the foundation for learning emotions.
Let’s use the emotion of “anger” as an example to explain the learning of emotions. When a person faces a situation and feels the emotion of “anger,” hormones, especially adrenaline, are secreted in response to the stimulus, and the sympathetic nervous system reacts.
This physiological response is remembered as “anger,” and the situation that triggered the stimulus is also remembered. Thus, the information that caused the emotion of ‘anger’ exists in the brain. By applying and analyzing this data, the brain determines whether to feel “anger” in response to a particular stimulus. Another piece of evidence comes from brain stimulation experiments.
When electrical stimulation was applied to the part of the brain responsible for laughter, the women who received the stimulation could not suppress their laughter. At the same time, they said, “You guys are really funny!” as if the doctors had made them laugh. This experiment shows that the brain can induce emotions by sending electrical signals to specific parts of the brain in certain situations.
In this way, the formation of emotions has many similarities with artificial intelligence and computer algorithms. Computers store data in memory devices, and artificial intelligence algorithms send specific stimuli to specific parts of the CPU. When a specific stimulus (command) is received, the computer retrieves the corresponding response from memory and executes it. If we interpret human emotions as responses based on experience and memory, then artificial intelligence receiving electrical stimulation in specific areas corresponding to specific emotional data values and expressing appropriate emotions is not significantly different from how humans express emotions.
One might wonder if the process of forming emotions, particularly primary emotions, is the key difference between computers and humans. Of course, artificial intelligence has not evolved from primitive organisms like mammals, developing a limbic system and cerebral cortex. Nor did it acquire primary emotions through evolution for survival and reproduction. However, what is important is the existence of primary emotions, not the process by which they were formed. Even if they are input by humans, as long as they possess primary emotions, their ability to collect and classify information about rapidly changing situations will render the limitations of human input meaningless. The technology that makes this possible is artificial synapses and deep learning technology.
Thoughts, emotions, moods, and memories are all regulated by neurotransmitters in the brain. The human brain has structural and functional units called neurons, which secrete neurotransmitters at their axon terminals. The area where a single neuron connects to tens of thousands of other neurons to exchange neurotransmitters is called a synapse. Neurotransmitters stored at the axon terminals of neurons are released when neurons receive neural information in the form of electrical signals, and these signals are transmitted to other neurons through synapses. Neurons that receive signals open sodium-potassium ion channels to form action potentials within the cell.
This spreads to surrounding cells, and the signal is transmitted. This is a very simplified representation of the process that regulates thoughts, emotions, moods, and memories, but it is the accumulation of these simple interactions between neurons that forms human thoughts and emotions. Artificial synapses are created by mimicking the human neuron-synapse system. This hardware, modeled after human neurons, has made it possible to create artificial intelligence that can respond instantly to any situation.
Previously, memory and input values were stored one-to-one, which limited the system’s response speed and diversity. However, artificial synapses, which mimic the process of a single neuron interacting with dozens of other neurons, not only overcame these limitations but also led to the development of deep learning technology. This marked the birth of a new hardware that resembles the human brain in both form and function.
The development and advancement of artificial synapse technology have driven the evolution of deep learning technology. Deep learning technology is used to cluster or classify objects or data. The core of deep learning technology lies in classification through pattern analysis. The amount of information that artificial intelligence processes is vast, and to analyze and classify this information efficiently, hardware more powerful than existing systems was required. This led to the development of artificial neural networks, which further advanced artificial synapses. Artificial neural networks use dozens of CPUs and GPUs (central processing units) to achieve fast and efficient classification. By leveraging GPUs for parallel processing in addition to CPUs for rapid information processing, these networks can classify astronomical amounts of data in a short amount of time. The reality where artificial intelligence stores data on various situations that evoke emotions, classifies them, and produces appropriate emotional expressions when faced with new situations is approaching. Classifying and predicting through pattern analysis, and connecting dozens of CPUs and GPUs into an artificial neural network is structurally and functionally very similar to the responses and structures between neurons in our brain and between the limbic system and the cerebral cortex. In fact, in March, scientists in France and the US announced that they had successfully developed a solid-state artificial synapse called a “memristor” for artificial brains that can learn on its own and even mimic the plasticity of human synapses.
Regarding this, it has been argued that human neuron synapses do not simply transmit signals but also regulate the strength of the transmitted signals, demonstrating highly variable performance, making it difficult to replicate with simple hardware. This raises the question of whether human biological characteristics can be completely replaced by mechanical components. However, even by examining the “memristor” mentioned above, we can see the positive potential. Memristors use dielectrics to change voltage and achieve plasticity. In this way, technological advances are leading to the development of new hardware that did not exist before. Although the components of this hardware are not identical to those of the human body, they perform the same structural and functional roles.
Emotions are created and expressed based on existing data on evolution and memory, similar to a computer retrieving stored data. Meanwhile, artificial synapses that mimic human synapses have been developed, advancing deep learning technology. Deep learning enables rapid classification of situations, allowing for immediate retrieval of appropriate responses. Combining these three elements, we can conclude that artificial intelligence possesses structural and functional units similar to the human brain, enabling it to process vast amounts of data and respond quickly and accurately to situations while expressing emotions. Although we have focused on emotions, the nonverbal, spatiotemporal, intuitive, and emotional functions performed by the right brain are also formed through the same process as emotions. Considering that all these functions originate from the interaction between synapses within the brain, we can extend the conclusion that artificial intelligence can possess emotions to the conclusion that artificial intelligence can perform the role of the right brain.
We discussed how the most representative emotion of the right brain was formed through evolution and how it appears in each situation, what the similarities are between the synapses of artificial intelligence and humans, and how artificial synapses have developed into artificial neural networks that can respond quickly to situations through rapid classification. In summary, emotions, which were born for survival and reproduction, remained in the brains of mammals and enabled them to know what emotions to express in what situations, and this is also possible for artificial intelligence. Furthermore, the development of artificial synapses modeled after those in the human brain has led to the advancement of deep learning technology. This enables artificial intelligence to quickly classify vast amounts of data and respond appropriately to situations in real time. Therefore, I agree with the claim that artificial intelligence can sufficiently replicate the role of the right brain.
The conclusion that artificial intelligence can also perform the role of the right brain has significant implications. If an artificial right brain (artificial intelligence) that can assist or replace the role of the human right brain is developed, the most anticipated social impact would be that it could provide a solution for patients suffering from right brain disorders. ADHD (attention deficit hyperactivity disorder), which has been showing an increasing incidence rate recently, is a representative right brain disorder. It is possible to mitigate symptoms by educating patients with artificial intelligence that performs the normal functions of the right brain or by implanting a chip equipped with artificial intelligence into the right brain to replace its functions. Additionally, there is potential for treating or alleviating conditions such as autism and non-verbal learning disorders. However, in such cases, it is essential to clearly define the boundaries between what constitutes human identity and what is artificial intelligence. In severe cases, if the brain is replaced with an artificial intelligence brain, there will be ongoing discussions about whether the patient should be considered human or a cyborg. Furthermore, if artificial intelligence with the functions of both the left and right brain and a physical body emerges, there will be debates about whether it should be classified as a new species.
