In this blog post, we explore how far artificial intelligence can catch up to human intelligence, examining its potential and limitations.
In his book ‘Sapiens’, Yuval Noah Harari warns that the end of Homo sapiens is approaching. One reason is that the development of cyborg technology will make humans no longer human-like. Indeed, the transformation from human to cyborg is already underway, driven by the scientific revolution. Currently, prosthetic limbs can replace parts of the body, and as bioengineering advances further, it will be possible to create replacements with performance rivaling that of biological organisms. For instance, an era is approaching where, instead of removing diseased tissue or transplanting organs from others, we will use artificially created replacements. However, significant debate persists regarding whether artificial intelligence can replace human intelligence.
Artificial intelligence has become indispensable in modern society. From the historic match between Lee Sedol and AlphaGo several years ago, to the current focus on self-driving cars, and household appliances advertising “artificial intelligence” features, AI is deeply woven into our daily lives. If this rapidly advancing AI were to possess human-like intelligence—that is, problem-solving capabilities—the day would come when AI could replace humans. Yet some deny the possibility of AI ever matching human intelligence. They argue AI cannot replace human intelligence primarily on two grounds. First, while AI might perform tasks like humans, it cannot understand what it is doing or the meaning behind those actions. Second, they claim AI cannot satisfactorily resolve issues related to values or ethics. However, these arguments are refutable, and the possibility of AI matching human intelligence remains entirely plausible.
The first argument that AI cannot replace human intelligence is that, unlike humans, AI lacks understanding of meaning. Humans can derive laws from phenomena or infer new facts, and they can apply given rules to create something new. Conversely, those who argue that AI cannot match human intelligence view it as merely complex software with predetermined outputs based on inputs. This argument is based on John Searle’s ‘Chinese Room’ experiment. In this experiment, a person who does not understand Chinese can still converse normally with a Chinese person outside the room by following a set script, yet that person does not comprehend Chinese. This can be interpreted to mean that even if AI can produce the correct output for a given input, it does not ‘understand’ the rules governing that process, and AI cannot use those rules to generate new outputs.
However, the ‘Chinese Room’ experiment may be a flawed analogy. Consider the human brain: individual neurons do not understand language, but the human brain, as a collective of neurons, does understand language. Similarly, even if the person inside the room does not understand Chinese, the room as a whole, viewed as a single system, can be said to understand Chinese. This demonstrates that artificial intelligence can possess the same level of understanding as humans and can become equivalent to human intelligence. Furthermore, by developing algorithms that identify problems or derive rules, artificial intelligence could understand the meaning of the problems it solves and develop an intent to address additional issues that arise. This refutes the claim that artificial intelligence can only perform predetermined tasks.
The second argument against AI replacing human intelligence is that AI cannot adequately address value and ethical issues. They contend that because AI cannot make value judgments like humans, replacing human intelligence with AI would cause numerous problems. Many problems in the world are tied to values and ethics. Since AI cannot solve problems without prioritizing specific situations, it is claimed that AI cannot replace human intelligence. This is because there are simply too many possible scenarios to define every situation.
However, AI has now reached a stage where it can learn on its own. Human ethical awareness and value judgment abilities are not products of evolution but rather the result of learning through diverse experiences from childhood. AI can learn like humans, and because it is a machine, it can experience even more situations. Therefore, the possibility of AI solving problems related to values and ethics is very high. Furthermore, while humans struggle with indecision and worry, leading to inconsistent value judgments, AI can make decisions in a very short time. This means AI has the potential to surpass human intelligence.
The claim that artificial intelligence cannot match human intelligence may have been persuasive in the past, but it lacks credibility today. Current artificial intelligence learns and draws conclusions by mimicking the structure of the human brain (neurons), such as through perceptrons. Claims that AI lacks understanding of the tasks it performs, or that it cannot match human intelligence, contain logical fallacies or stem from a lack of understanding of modern technology and strong AI. This essay has argued that AI possesses human-level understanding and can reach the same level as human intelligence. Therefore, the possibility that AI could replace human intelligence is entirely plausible. However, the issue of AI replacing human intelligence still requires considerable deliberation. Even if humans become less human-like, the utilization of AI should only occur after ethical and institutional developments have been achieved. As Yuval Noah Harari mentioned in ‘Sapiens’, we must ask ourselves the question of what we want to become.
