Generative Designer Tools

Integrating AI into the Repertoire of the Designer



Abstract:

The proposed research seeks to explore the use of AI generative processes as a new design-methodology. Unlike traditional methods that rely on manual tools or parametric-design coding, this approach leverages massive AI models that encompass vast amounts of image data. By embracing a playful strategy: encompassing randomness, experimentation, distortion, and selection, we aim to establish better interaction with these deep learning systems, leading to unexpected and novel outcomes.k

The study will delve into the understanding of the concept of latent space, where AI models represent images as arrangements of pixels in a mathematical space. The latent space captures meaningful clusters and variables that separate different objects and features within the image dataset. By learning to understand how to navigate through this multidimensional space, creators can prompt the AI models to generate specific design studies, based on their desired criteria, such as shape, texture, or structure.

Nevertheless, the research must acknowledge the challenges associated with AI-generations. The latent space contains a hidden dimension of associations learned from its internet-based datasets, which can perpetuate societal biases and limitations.

To unlock new possibilities for design-through-AI integration, the research will combine theoretical analysis, experimental investigations, and case studies. By understanding and refining the prompt engineering process and exploring the latent space, creators can harness the power of AI to enhance their personal creative process, redefine boundaries, and potentially shape the future of design thinking.

Keywords: Multidimensional, Latent Spaces, Compositional Affects, Prompt Playing, Looping Problems, Purifiers, deep learning models, generative process, diffusion, distortion, compression, mirror of society, biases in datasets, compilations, transformers, infinite possibilities, fine-tuning, curation, originality.

Marcel Moonen 11.11.2023, Generative Designer Tools  [EDUCATIONAL PURPOSE ONLY] Triple-A Society, M. Production