In Homo Deus, Yuval Noah Harari explores the relationship between humans and artificial intelligence, and the possibility of AI surpassing humans in all areas. He asks whether consciousness will preserve our uniqueness, or whether it will eventually render us obsolete. In response, I argue that consciousness may not be what makes us special, and that artificial intelligence may eventually overtake all human abilities.
Yuval Noah Harari’s “Homo Deus: A Brief History of Tomorrow” is a book about what our fate will be in the future, and he has a very interesting perspective on how the future will unfold. Harari interprets all life, including humans, as nothing more than algorithms, and argues that while humans currently dominate the planet, AI’s capabilities are catching up to us. In the future, he explains, AI’s intelligence will surpass human intelligence in every field.
At the end of the book, the author asks the question: “What is more valuable, intelligence or consciousness?” After all, one could argue that consciousness, which exists in humans but not in A.I., may be what makes us special.
But I don’t think consciousness is what makes us special. This is based on two arguments. First, we can’t really determine if an entity has consciousness, we only care about physically interacting with it. Second, AI can act as if it has consciousness. If both of these arguments are true, then we can say that A.I. is capable of being equal to humans in terms of consciousness.
First, what is consciousness? In this book, we define consciousness as a kind of subjective experience. It’s an experience that you can know about, but that no one else outside of yourself can verify. We know that consciousness is caused by electrochemical reactions in the brain, and we know that these reactions perform some data processing function. According to current conventional wisdom, humans are conscious. It is also believed that higher animals such as dogs and cats have consciousness. On the other hand, it is debatable whether ants, plants, bacteria, etc. have consciousness. It’s generally believed that things like computers and smartphones don’t have consciousness.
But how do we know that an entity has consciousness? Consciousness is a subjective experience, but can it really be known? Historically, people have looked for the presence of features of consciousness, such as sensations or desires, to determine whether an entity is conscious, but this is not enough to say that it is conscious. These features could be the output of a non-organic algorithm. In the end, we can’t strictly determine if the entity is aware of itself and its experiences. The same is true when we interact with animals. When we own a dog or cat, we treat them as if they are conscious. But we don’t actually know if they really are conscious, we can only infer inductively from their behavior or facial expressions that “they are conscious!”
This logic is not meant to explain that animals don’t have consciousness. It’s to explain that even if they do have consciousness, in interactions between individuals, we can only see the surface (behavior, facial expressions, etc.) of their supposed consciousness. We only see the surface and say, “This thing has consciousness.”
So if an A.I. can accurately represent only the surface, we can look at it and say, “It has consciousness,” right? Yuval Noah Harari seems to think that AI can surpass humans in terms of intelligence, but not in terms of consciousness. His reasoning is that consciousness and intelligence are two different things, and that current AI has only advanced intellectually, not consciously at all.
But I think this is a matter of technology. It’s not yet possible to make a single A.I. perform multiple tasks as well as a human. We’ve seen some AIs that are better than humans in certain areas, such as Go, chess, and object recognition, but not in all areas. This is the “weak AI” stage. It’s possible that this specificity of AI is why computers are currently not considered conscious. But one day, AI will surpass human intelligence in all areas. This kind of general-purpose AI is called “strong AI”. This has yet to be realized. However, most scientists believe it will be possible within a few decades. The brain is very complex, but it doesn’t violate any of the physical and chemical laws we know, so it should be possible to build a strong AI.
Of course, human-level intelligence doesn’t necessarily mean human-level consciousness. As Yuval Noah Harari says, there is a gap between intelligence and consciousness. However, if there are beings with human-level intelligence, I believe that consciousness can exist. For example, when you first design an A.I., you give it a goal: “When interacting with humans, make it as undetectable as possible that you are not human!” A strong AI should be able to fulfill this goal. If it does this well, can it really be called conscious? You might argue that an A.I. is not conscious because it doesn’t actually understand itself or its opponent, even if it is superficially similar to humans. However, there’s nothing wrong with that. It’s like Samantha, the AI in the movie Her. In reality, Samantha is just an algorithm, but to people, she appears to have consciousness. If the technology is good enough to create a strong A.I., there will definitely be an A.I. like Samantha.
In conclusion, the comparison between humans and A.I. is a comparison of how they react to a situation (input) and what they do (output). In this respect, human consciousness itself is practically irrelevant, and I believe that AI will eventually overtake all human abilities. If this happens, will most humans eventually become redundant, and in the words of Yuval Noah Harari, will we become mere ripples in a cosmic-scale data flow? Or will we recognize the dangers of strong AI and stop developing it? No one knows what will happen. But what we do know is that we need to prepare for the coming development of AI and continue to explore the meaning of humanity’s existence.
