The Cage of Zeus:
In the novel, ordinary humans consider the Rounds to be abnormal, while Tei, as a Round, in turn considers ordinary humans' concepts of gender to be abnormal. I think this made me think on society's definition of "normal." Many of the social classifications we take for granted are not all directly stipulated by nature, but are formed within specific historical, cultural, and social environments. Of course, this is not to say that there are no objective biological differences between men and women, but rather that the meanings we assign to these differences and what kinds of bodies are considered normal or ideal are all influenced by social concepts. Further, from Tei's story, it can be seen that even if you create a new society in which everyone has the same reproductive structure, differences may still emerge between individuals, and society may still build new standards of normality around these differences.
Self-reference Engine:
I'm a data science major, and reading this novel first made me think that it is saying that having information is not the same as understanding information. When we train machine learning models now, there is usually a basic idea: increasing the amount of data, model capacity, and computational resources can improve performance on many tasks. For example, training a language model: we collect a large amount of text; the model learns the statistical patterns in the text through training; the model learns the relationships between words, grammatical structures, and context; ultimately the model can generate text that conforms to linguistic patterns. But the novel constructs an extreme situation. Suppose every symbol is unique and there are no known rules connecting them. Then we cannot simply infer language structure through statistical repetition frequencies. So in this case, it is saying that more data is useless for the model.
On the other hand, although this is a bit out of the topic, the novel writes that this Giant Corpora of Knowledge is essentially also a giant neural network model, that is, its underlying logic is the same as what we now call AI (GPT, Claude). So can such an apparently omniscient and omnipotent AI truly count as an agent that can understand problems or understand words? I don't want to discuss technical details too much, but to summarize in one sentence: the current neural network structure AI does not understand your question at all, but it is just a complex context prediction machine, that is, it generates the most likely continuation based on your input. So this makes me wonder: if we are truly pursuing an agent that can understand problems or sentences like a human, then is our current AI research somewhat off track?
First Rate Material:
I think this novel, like The Cage of Zeus, also discusses the issue of social construction. But what interests me more in this novel is the character Naoki. In the first half of the story, Naoki seems like a person with very strong principles. Even though the whole society considers the use of human body materials to be correct, he still insists on his own position. But when Naoki discovers that his father's skin has been made into a wedding veil, his values suddenly begin to collapse. Because in his original understanding, human body materials only meant death, corpses, and disrespect for the dead. But now this veil suddenly has an additional layer of meaning for him. It is no longer just human skin. It is a gift left by his father, a memory, and the emotion between the two of them. This made me think about our moral judgments — that is, perhaps sometimes people with such firm moral positions are all through purely rational derivation, yet they ignore that emotion may also be an important basis for moral judgment. For example, Naoki previously thought he had grasped complete moral principles, but in reality he had never truly faced complex real situations. Because real situations are full of a mixture of rationality and emotion, and often we may not be able to fully follow that firm, rationally derived moral position.
No comments:
Post a Comment