Neural networks represent the cutting edge in artificial intelligence, revolutionizing the way machines learn, interpret, and generate data. These intricate systems consist of countless interconnected “neurons,” each responsible for processing specific portions of input before channeling their outputs through the expansive network. This architecture underpins recent breakthroughs in universal text and image generation models, which are capable of producing sophisticated, high-quality outputs from vast data sets. Despite their impressive performance, neural networks present challenges, including intensive data requirements and limited adaptability beyond their training scope. Additionally, their immense complexity renders the precise decision-making processes largely inscrutable, as the influence of thousands of parameters on any given outcome remains opaque.
One particularly fascinating application of neural networks is in AI-powered art generation. These systems translate textual prompts into vivid visual representations, blurring the boundary between human creativity and machine computation. Artists like Sougwen Chung have embraced these technologies, leveraging neural networks as collaborative partners rather than mere tools. Chung’s pioneering work in the arena of AI and robotics harnesses machine learning to augment her drawing and design capabilities, creating a novel intersection between human intuition and algorithmic precision. By situating AI as an active co-creator, her projects push the envelope of what it means to collaborate with technology in the realm of art.
Chung’s ongoing series, titled Drawing Operations Unit, spanning versions 1 through 4 (D.O.U.G._1–4), exemplifies this symbiotic dynamic. Initially, these AI programs were trained on a dataset meticulously compiled from two decades of Chung’s own artistic output. This approach allowed the neural network to internalize and replicate her distinctive style, mirroring its nuances with remarkable fidelity. The early iterations of D.O.U.G focus on mimicry—teaching the AI to draw in a way that echoes its human predecessor. However, as the series progressed, the project evolved into something far more complex and interactive.
For instance, D.O.U.G._2 was employed in live performances where it was physically embodied as a robotic hand. This robotic extension enabled the AI to draw alongside Chung in real time, transforming the robotic hand into a tangible co-creator, rather than a mere virtual programmed entity. This material manifestation of AI radically redefined the notions of authorial control and artistic agency, showcasing a layered collaboration between human spontaneity and algorithmic response. Audiences witnessed an unprecedented dance between artist and machine, where the boundaries of machine creativity were explored in real-time artistic dialogue.
The subsequent iteration, D.O.U.G._3, ventured into collective creation by integrating environmental data from New York City into its drawing process. Using motion vectors extracted from public surveillance cameras, the AI was programmed to interpret the flow of pedestrians and vehicular traffic as a source of creative stimuli. This integration effectively transformed the artwork into a dynamic living entity that responded to and was shaped by the pulse of urban life. One notable demonstration of this concept was in Chung’s 2018 work “Omnia per Omnia,” where she collaborated with five small robotic actors powered by D.O.U.G._3. The robots and Chung painted together on a massive canvas laid on the floor, merging human artistry with ambient urban movement in a groundbreaking urban-robotic symbiosis.
This experiment showcased another groundbreaking facet of AI-human collaboration: the extension of the creative process beyond the individual to include collective, environmental inputs. By incorporating real-world movement data, Chung’s project highlighted the potential for AI to mediate and even amplify the creative energy circulating within an entire cityscape. Such projects challenge traditional concepts of singular artistic authorship and invite reconsideration of art as an evolving, distributed experience that transcends geographical and individual limits.
Despite these advances, the nature of neural networks imparts intrinsic constraints. While D.O.U.G shows evidence of learning and creative adaptation within the domain of visual art, it remains narrow in its applicability. Neural networks exhibit a data-hungry dependence and are largely limited by the scope of their training. They lack the fluid adaptability and contextual awareness characteristic of biological organisms. Moreover, the internal reasoning pathways of these networks are notoriously difficult to dissect or interpret, which limits our understanding of how they arrive at creative conclusions. Their “black box” character stands in stark contrast to the transparency human cognition often affords in social engagement.
This obscurity challenges traditional notions of shared cognitive processes and has prompted some artists and researchers to seek alternatives for constructing interactive AI art systems with the semblance of sentience or intentionality. One promising direction is Inductive Logic Programming (ILP), an AI paradigm that leverages symbolic reasoning and logic to generate transparent and interpretable decision pathways. Unlike neural networks, ILP offers clearer representations of its internal logic, facilitating more relatable and socially engaging interactions with human users.
The capacity of AI art generators like D.O.U.G to serve as independent yet collaborative agents offers profound implications for the future of creative expression. By positioning AI as partners rather than mere tools, these systems invite us to rethink the nature of artistic agency and creativity. The collaboration fosters an experimental mode of human-technology symbiosis, whereby machines contribute not only technical skills but also unique creative impulses derived from algorithmic learning. This reconceptualization could redefine artistic practices, broadening the spectrum of who or what can be considered an artist in the digital age.
Yet it is crucial to appreciate the distinctions between AI-generated creativity and human cognition. While AI can emulate aspects of artistic style and adapt within defined parameters, it does not possess consciousness, intentionality, or the holistic experiential understanding that informs human artistry. Recognizing these limits is important for framing AI as a creative assistant rather than a replacement for human creativity. The enigmatic, opaque decision-making of neural networks likewise underscores the challenges in fostering genuine social engagement with AI entities.
Ultimately, innovations like Chung’s D.O.U.G project exemplify how AI can enrich human creativity by introducing novel modes of interaction and collaboration. These projects probe new artistic frontiers, demonstrating how neural networks can transcend passive tool roles to become active partners shaping the creative process. The infusion of environmental data and robotic embodiment further expands the possibilities for dynamic, responsive art forms. This emergent landscape of AI art signals a fertile ground for both artistic and technological experimentation.
The exploration of AI-human collaboration in creative contexts prompts critical reflection on authorship, creativity, and technological agency. It illuminates the complex interplay between algorithmic processes and human intuition and invites dialogue on the evolving definitions of art and artisthood. Neural networks serve as fascinating vehicles for this exploration, showcasing unprecedented capabilities while simultaneously highlighting the ethical and philosophical questions raised by increasingly autonomous creative systems.
Looking ahead, ongoing research and artistic experimentation will likely deepen our understanding of how AI can complement and extend human creativity. Interdisciplinary approaches, blending computer science, robotics, cognitive science, and art, will be paramount in crafting systems that are not only powerful but also interpretable, adaptable, and socially responsive. As AI art evolves, it holds the promise to reshape creative landscapes, democratizing access to new tools and challenging long-held assumptions about the sources of artistic agency.
In the face of such transformations, artists like Sougwen Chung stand at the vanguard, demonstrating how AI’s enigmatic capabilities can be harnessed with sensitivity and vision. Their work operates at the intersection of technology and humanity, raising profound questions about collaboration, interpretation, and the future of creative practice. Neural networks and interactive AI are no longer mere curiosities but foundational elements shaping contemporary artistic expression and the broader relationship between humans and machines.
Subject of Research: Exploration of neural networks as interactive mediums in AI-driven art and their collaborative potential with human artists.
Article Title: Art with agency: artificial intelligence as an interactive medium.
Article References:
Sklar, S.J., Jiang, M. Art with agency: artificial intelligence as an interactive medium. Humanit Soc Sci Commun 12, 1546 (2025). https://doi.org/10.1057/s41599-025-05863-z
Image Credits: AI Generated
