In this blog post, we will look at what technologies and elements are needed to implement strong artificial intelligence with autonomy, emotion, and reason.
From driverless cars, AlphaGo, and fingerprint recognition devices to robots that think and act on their own, which can be seen in movies, they are all machines based on artificial intelligence. As artificial intelligence (AI) technology is rapidly advancing, it is deeply embedded in various fields of modern society, and its application areas are becoming increasingly diverse. Various smart devices and automated systems used in our daily lives are designed based on artificial intelligence, and the purpose is to make human life more convenient. However, a fully autonomous robot that thinks and acts on its own, as in the movie, is very different from the artificial intelligence technology we currently have. This is because real-life artificial intelligence is mainly limited to processing set data and performing given tasks. So, what elements need to be added to create a robot that acts on its own from a simple machine? To understand this, we need to take a closer look at the types of artificial intelligence and the direction of technological development.
First, let’s take a look at the difference between weak and strong artificial intelligence, which is the biggest difference in the stages of artificial intelligence development. The fingerprint recognition machine and the self-driving car mentioned earlier belong to weak artificial intelligence. Weak artificial intelligence refers to AI that has no self and can only act as programmed by humans. It is an algorithm designed for a specific purpose and is only used to perform specific tasks. For example, AlphaGo, which we are all familiar with, is also a weak artificial intelligence and was designed with an algorithm that is specialized for playing Go. It can perform exceptionally well in a specific task called Go, but it cannot be used in any other situation. As such, weak artificial intelligence is optimized for specific tasks and is used only for a limited purpose, just like a tool.
Drug-based AI is subdivided into general drug-based AI and universal drug-based AI. General drug-based AI simply does one thing in a specific field. For example, tasks such as distinguishing spam emails or finding cancer cells in medical images are examples of general drug-based AI. On the other hand, universal drug-based AI is a more advanced form that has an added learning function. As a result, people can learn to cook a dish they have never cooked before by watching someone else do it and then follow along, or they can take over and perform a simple task. General-purpose drug AI can be said to be in the early stages of transitioning from drug AI to strong AI. At this stage, machine learning and deep learning technologies are important because they require the ability to learn new information and respond to situations beyond simple repetitive tasks.
In contrast, the Terminator, the robot in I, Robot, and Jarvis from Iron Man, which we often see in movies, are strong artificial intelligence that act with self-awareness. Strong artificial intelligence refers to artificial intelligence that has a self and can make its own judgments and take action. The goal of strong artificial intelligence is to have an intelligence level similar to that of a human. Such artificial intelligence, which includes not only reason but also emotion, has autonomy and can make decisions and respond to situations on its own without specific instructions. The biggest difference between strong artificial intelligence and weak artificial intelligence is the presence or absence of self. If there is a self, the machine can act based on its own purpose and judgment.
Developing artificial intelligence requires a level of reasoning and emotion similar to that of humans. Many people do not question whether artificial intelligence has reasoning, but there is much controversy over whether it needs emotion. Reasoning is essential for analyzing and making decisions based on given data, but emotion is a necessary element for understanding complex social contexts and emotional situations of humans beyond simple analysis. Current AI is gradually acquiring the ability to analyze and make decisions based on data, but it is still difficult to imitate complex emotional states at the human level. For example, AI such as chatbots and virtual assistants can mimic basic emotions, but they only show predetermined reactions rather than actually feeling emotions. However, there is an opinion that both of these elements are necessary to develop strong artificial intelligence because reason and emotion interact to make better decisions.
The need and possibility of developing strong artificial intelligence is being discussed in academia and industry. If we develop strong artificial intelligence, it has the potential to bring about major changes in human society. In particular, the help of artificial intelligence, which can make sophisticated decisions that go beyond human judgment in the fields of medicine, education, law, and the environment, is expected to have great social value. For example, in an aging society, artificial intelligence can improve the quality of medical services by making medical decisions and diagnoses, which will increase the accuracy and efficiency of medical care. In addition, if artificial intelligence analyzes and makes decisions to solve environmental problems, it can greatly contribute to moving toward a sustainable future.
In order to develop weak artificial intelligence into strong artificial intelligence, we must first develop the rational judgment ability of weak artificial intelligence to a much higher level. Expert systems process large amounts of data to help us understand various situations based on conditions, but they are still far from human high-level thinking. Machine learning, or machine learning technology that enables computers to analyze new data by creating their own rules, gives computers more flexible thinking abilities. In addition, genetic algorithms can also be a powerful tool. Genetic algorithms do not focus on finding a specific answer, but on finding the best answer among possible alternatives, and they mimic the way humans adapt to various situations.
Giving artificial intelligence emotions is also an important task. Human emotions are very complex and subjective, so there are limitations to simply programming them to respond to certain emotions depending on the situation. Currently, emotions are simulated in a way that allows them to mimic emotions determined by a specific situation, but this is not considered a true expression of emotion. To develop AI with emotions, we need to understand the human emotion system more deeply and study models that reflect various elements of emotion. When emotions are given to AI, it can elicit more human-like responses, which makes communication between humans and AI more natural.
If we give reason and emotion to weak artificial intelligence in the above way, we will be able to create strong artificial intelligence someday. If this technology is perfected, strong artificial intelligence will go beyond being a mere tool and become a partner that can solve various problems in cooperation with humans. With the advancement of technology in the future, strong artificial intelligence is likely to become a reality rather than a story of the distant future.
