The Hidden Layers
Alternative Machine
Concept Design
This project explores the ”layer structure” of machines when they process language. I designed an alternative machine that challenges our assumptions of language by visualizing how machines decompose and reconstruct meaning through hidden layers.
Humans understand a word holistically, we immediately associate it with images, emotions, and layered meanings. Machines, however, do not perceive language as a unified whole. Instead, a word is decomposed into tokens, classified by intent, and interpreted through contextual inference. These processes occur within hidden layers invisible to the user. This project used Interactive recognition and graphic archive explores the possibility of treating language as visual input, proposing new perceptual approaches to language while examining the fundamental differences between human semantics and machine semantics.
解构机器
概念设计
本项目探索机器处理语言时的“层级结构”。我设计了一种另类机器，它通过可视化机器如何通过隐藏层级分解和重构意义，挑战我们对语言的固有认知。
人类对词语的理解是整体性的，我们会立即将其与图像、情感和多层次的含义联系起来。然而，机器并不将语言视为一个统一的整体。相反，一个词语会被分解成词元，根据意图进行分类，并通过上下文推理进行解读。这些过程发生在用户不可见的隐藏层级中。本项目运用交互式识别和图形档案技术，探索将语言视为视觉输入的可能性，提出新的语言感知方法，并考察人类语义和机器语义之间的根本差异。
