Artificial intelligence, which emerged with the development of computers, is currently providing convenience to humans in various fields. AI is divided into weak and strong AI, and while the implementation of strong AI is still considered technologically impossible, the possibility of its development is open. With the combination of technologies such as deep learning and big data, AI’s ability to learn is getting stronger and stronger, and it will only get better in the future.
When computers first appeared, their role was to perform many calculations that humans could not realistically perform. Since then, as computer technology has evolved, computers have helped humans in many areas, including basic calculations, storing and sharing information, and complex computations. Computer science is developing at the fastest rate compared to all other fields. Currently, people are working on making computers not only do things that humans can’t do, but also think and learn on their own, something that only humans and higher organisms have been able to do before. This effort to enable computers to think like humans and to learn and develop on their own has resulted in a technology called artificial intelligence.
Artificial intelligence is not just AlphaGo or Watson, as most people think of them, but any technology that enables computers to think and learn things that only humans can do. For example, automatic translation technology doesn’t just translate a list of words in order, but also changes the meaning of words to make sense in context. Recognizing a person’s face in a camera or categorizing photos of the same person are also examples of AI. These technologies are used in a wide range of fields, and they have recently become a part of our daily lives.
AI can be categorized into two main types: weak and strong AI. Weak A.I. refers to A.I. that is designed to solve problems with a specific purpose. All AI technologies developed to date belong to the weak AI category. On the other hand, strong A.I. is an A.I. that can learn and solve all fields by itself without a specific purpose, and it is widely believed that it cannot be created with current technology. An example of a strong AI is Skynet from the movie The Terminator. The key to strong A.I. is that it can make its own judgments, learn, and come up with means to achieve any goal. AlphaGo is often confused with strong AI. This is because AlphaGo has great computational power and the ability to learn numbers on its own. However, AlphaGo is categorized as a weak AI because it finds multiple solutions with the specific goal of winning the game of Go.
The closest thing to a strong AI program developed to date is IBM’s Watson. Watson has gained a lot of recognition for its performance by participating in and winning popular quiz shows in the U.S. It is also expected to have gained more knowledge from competing against trivia masters. And Watson wasn’t just built for quiz shows. Watson can work with systems in many other fields and learn from them as well, making it a very useful artificial intelligence that is rapidly becoming more versatile. Of course, Watson is not yet a strong AI because it is given a purpose by humans and operates to fulfill it, rather than finding solutions in various fields on its own. However, with the accelerating pace of computer technology development, combined with technologies such as deep learning and big data, the possibility of a strong AI is open.
As mentioned earlier, A.I. eventually learns to improve itself, and a technology called machine learning enables it to learn. Machine learning is a technique that allows computers to learn without explicitly written programs. AI with machine learning analyzes the relationships between the data provided to it, identifies unspecified mechanisms, and produces new data. In simple terms, it is a technology that analyzes many A’s and B’s to understand the process of how A leads to B through data. Machine learning is about using these patterns to generate B from A. This technology is the basis for A.I. to learn itself, and recently, deep learning technology was born from it, and deep learning technology has played an important role in the rapid development of A.I.. In traditional machine learning, the algorithm is trained by humans to discover and produce answers based on information provided by humans. Deep learning, on the other hand, has the ability to find answers on its own, without the need for humans to teach it a lot of information. For example, with machine learning, in order for an AI to look at a picture and classify it as a bird, a human would have to show it several pictures of birds and teach it “this is a bird.” Deep learning, on the other hand, can analyze different pictures of birds and classify other birds as birds without human instruction. Thus, deep learning makes AI much more efficient and capable of learning on its own in a variety of areas.
In order for deep learning to work well, AI needs a huge amount of data available to it. This is where the aforementioned big data technology comes into play. Big data technology is a technology that processes data that cannot be handled by conventional processing methods due to its large volume and short generation cycle. Deep learning technology is used in combination with big data because the more data available, the more effective and accurate learning is possible. In this case, the way we process and use the growing amount of data will also grow through learning. By combining deep learning and big data technologies, AI is able to learn and utilize larger amounts of information more efficiently.
Most researchers believe that strong AI is not possible with current technology, but AI technologies are evolving, including the shift from machine learning to deep learning and the combination with big data technologies. Through the interconnection of these technologies, AI is evolving at a rapid pace, and this pace is likely to accelerate. In addition, humans will continue to strive to improve AI. Therefore, even if it is impossible to create a strong A.I., it will continue to evolve to get closer to it.
