Is a strong AI possible? How will strong AI affect humanity?

Let’s take a look at the future of artificial intelligence and see if it will ever threaten the existence of the human race.

 

Iron Man, Interstellar, The Matrix, The Terminator. These movies all have one thing in common: artificial intelligence. In movies, AI is different from the machines and objects we usually encounter. For example, the computer I’m using to do my assignment doesn’t understand me or move on its own. It only performs functions when I give it commands using a keyboard or mouse. In movies, however, artificial intelligence understands human speech and makes its own decisions. In Iron Man and Interstellar, the AI program Jarvis and the robot T’Challa act as assistants, carrying out various commands for the protagonist. On the other hand, the AI robots in The Matrix and The Terminator try to destroy human civilization and even control the entire human race. Of course, this is a level of technology that’s impossible today. It only exists in movies. However, we can’t ignore the possibility of future research. Could there ever come a time when the existence of humanity is threatened by artificial intelligence?
First, let’s take a look at what artificial intelligence is. Artificial intelligence can be divided into ‘artificial’ and ‘intelligence’. “Artificial” is easy to understand. It refers to things that are not naturally occurring, i.e., created by humans. However, the meaning of “intelligence” is harder to define. The dictionary definition is “the ability to engage in intellectual activity,” but many scholars have different definitions of its elements, scope, and characteristics. However, most agree that it is the ability to utilize previous experience, adapt to new objects or situations, and learn. This is not too different from the meaning of intelligence in general: AI is something (a robot or program) that was created by humans and has intelligence, that is, the ability to adapt and learn new things. Depending on the scope of its intelligence, AI can be categorized into weak and strong AI.
Weak AI is AI that functions as if it has intelligence only on the surface. Almost all AI currently in use and being developed, such as programs in self-driving cars, programs that search the internet for you through voice recognition, and programs that can compete with you in simulation games, are included in weak AI. They appear to be intelligent, but they are limited in that they can only make judgments about things that have already been fed into the program. They can’t adapt to new situations and objects. Self-driving cars may appear to adapt to new situations because they recognize and judge objects or people that are new to the road or that suddenly enter their path and stop or detour. However, the criteria for recognition and judgment cannot be outside the scope of the developer’s input, and it is not possible to utilize empirical knowledge such as “I hit that yesterday, it was a person, I should avoid it today. Therefore, it cannot be said to have intelligence. Of course, there are programs that can learn. There are also programs that change their judgment criteria by continuously synthesizing the events that occur, but the method of changing the criteria has already been entered by the developer and cannot be deviated from. Therefore, any A.I. that operates as intended by the developer is bound to be a weak A.I..
A strong A.I., unlike a weak A.I., is one that has real intelligence, not just apparent intelligence. It is an AI that can think and act like a human, or think and act beyond a human. It can adapt, learn, and think in completely new situations and objects. It is self-aware and can set goals. The most essential part of a strong AI is the phenomenon of emergence, which is the ability to create something new. In other words, a strong AI is almost no different from a human being. Therefore, a strong AI is an AI with a mind. While programs or robots with minds do not currently exist, there is much debate about whether they can be developed as technology advances, or whether they could emerge by chance.
There are two main approaches to strong AI. One is through programming algorithms, and the other is through biotechnology. Developing AI through computer programming is the first and most accessible approach, as it’s the one that’s still being done today. However, this is not the only way to achieve strong AI. Professor Dae-Sik Kim (2015), a brain scientist at KAIST, said in a lecture at Ewha Womans University that the resolution of language is lower than the resolution of thought. We cannot express all of our thoughts in language, so it is impossible to create a program by inputting human thoughts into a machine’s language when human thoughts cannot be expressed in human language. For example, if a program has the criterion that “cats have fur,” it will recognize a dog with fur as a cat. And it would recognize a cat without fur as not being a cat. Of course, you also need a criterion for “fur”. In fact, if you add more criteria, it can recognize cats. However, as currently implemented, the technology doesn’t actually identify a cat, but rather calculates the probability that it is a cat. This is because too many criteria can lead to objects that are not cats being recognized as cats, and too few criteria can lead to objects that are not cats being recognized as cats. The abstract boundaries that separate things cannot be clearly expressed in human language, and therefore cannot be programmed.
The second, biotechnological approach, is an attempt to create a human brain. It’s an attempt to artificially create the neurons that make up the human brain. Lulu Qian’s team at Caltech (2011) created four artificial neurons from 112 strands of DNA. The system can reportedly distinguish between people with training. Of course, the human brain consists of more than 15 billion neurons, just on the surface of the cerebrum, so it’s not yet comparable. But there’s nothing to say that we can’t build more than 15 billion artificial neurons in the future. Synapses, the connections between neurons, are also known to change over time, and if we can fully understand the mechanisms of these changes, we will be able to artificially create a human brain, although it may take a few years. This leaves us with one final question. Does a mind exist in a brain? If a mind exists in an artificially created brain, it follows that strong artificial intelligence is possible.
It is possible to prove this experimentally if only an artificial brain is implemented. If using humans as test subjects is an ethical issue, we could use long-haired animals that can use other tools. However, discussing this without experimentation is beyond the scope of science or engineering. There have been similar experiments in the past. In 1970, Robert White transplanted a monkey’s brain and spinal cord into the body of another monkey, which survived for nine days before dying of immune rejection. During the survival period, the monkey behaved similarly to before, which is problematic because it’s not an artificial brain. However, if an artificial brain were to be implemented, similar results would be expected.
In conclusion, strong artificial intelligence depends on the development of artificial brains. If an artificial brain is developed and a mind exists in the brain, strong AI can be realized. However, whether this strong AI will pose a threat to human existence is another question. We don’t see cows or monkeys as a threat to human existence. And we know how weak and compliant they can be, even in the face of absurdity, due to brainwashing. Therefore, the emergence of sentient A.I. is unlikely to be beyond human control.