This project investigates the boundary between human and artificial intelligence, conceptualising it as a condition related to information transmission and media structures. Human cognition is founded upon the continuous electrochemical reactions within the nervous system. Information travels along neural connections, with pathways and intensities undergoing dynamic shifts. Artificial intelligence, conversely, operates through discrete steps, repeatedly computing complex mathematical functions. Its information processing is confined within predefined parameters and update rules.

To translate this distinction into a perceptible structural relationship, I have chosen to present it using a deliberately simple physical setup: a 60×50×8-millimetre foam board into which five different sizes of nails are embedded. The foam board is spray-painted a silver-grey close to the nails' hue, thereby minimising material and colour variations. This allows the viewer's attention to focus more readily on the inherent rules of the structure itself.

On one side of the installation, nails of uniform size are arranged in a regular pattern, their overall form resembling a 'height map' generated within a uniform computational space. Localised shifts and band-like variations correspond to the 'computational boundaries' formed by AI during discrete computations. The opposite side employs nails of varying sizes, adopting a structure more akin to neuronal morphology: taller nails resemble neuronal bodies, while shorter, extended pathways mimic axons, exhibiting pronounced path-dependent characteristics. The two structures are not rigidly separated by a neutral dividing line, but rather meet through a zone of gradual, overlapping transitions. I intended for the boundary itself to function as a condition for continuously generating information, rather than a demarcation.

This project's research methodology integrates artistic practice references, neuroscience theory, and analysis of AI computational models. Formally, I drew upon Günther Uecker's approach of constructing spatial order through repetitive, dense, and non-narrative elements. As a highly standardised industrial object, the nail proved ideal for building systems governed by pure rules, avoiding symbolic interference and enabling viewers to directly perceive how boundaries manifest through distribution and density. In understanding the human nervous system, I drew upon Rall's 1959 Cable Theory, which treats neuronal dendrites and axons as a continuous medium to describe the propagation and integration of electrical signals through space—providing an excellent model for comprehending the boundaries of continuous information transmission.

Regarding AI, my approach primarily follows the framework of conceptualising neural networks as function approximation systems. From this perspective, AI's ‘thinking’ can be viewed as a process of numerical evaluation and parameter updating for high-dimensional continuous functions across discrete time steps (referencing Goodfellow, Bengio & Courville, Deep Learning, 2016).