Steve, A Framework For Augmenting The Visual Identity Design Process With ML

This research is positioned in the field of graphic design and seeks to investigate the working processes in visual identity projects and their augmentation through Machine Learning (ML). It defines identity as the visual elements that, together, create an atmosphere around a client, involving its values and views of the world and society. Through a deep focus on the creative process, this thesis proposes functional approaches to integrate the designer's perspective on the development of new digital tools. My study reveals fruitful ways to augment identity design through ML rather than replace designers through automation. Since its blooming during the Industrial Revolution, visual identity remains the highest-order project in the discipline of graphic design. The parallel evolution of graphic and information technology has undergone numerous phases in which visual identity structures have become more dynamic, and its impact on society has grown along with the designer’s responsibilities. Increasing integration of automation into graphic design in the twenty-first century, as well as potential future developments in ML, represent new challenges for professionals and researchers. Investigation into the intersection of ML and graphic design has been led mainly by computer scientists, leading to misplaced assumptions of creativity. At the same time, research into graphic creative processes is limited. My research addresses these deficiencies, and the gap in the existing literature on the conjunction between graphic design theory and practice, by involving practitioners in the evaluation and proposal of novel design tools. Moreover, it creates a direct link between software development and the actual needs of graphic designers. The novelty of this research lies in the intersection of design methodology, visual identity and ML. Research on design processes is well established in other areas like architecture, industrial design and software development. An understanding of tools and concepts from these fields helps to investigate the possibilities of integrating ML into the design process. Three main questions are addressed in the research: – Is it possible to find coherent working methods in visual identity projects? – What are the most critical phases for the designers in visual identity projects? – How can these be augmented through ML? To answer these questions, I utilize grounded theory methodology, complemented by literature review, to construct a conceptual framework rooted in the expertise of practitioners. By conducting semi-structured interviews with a sample of twenty graphic design studios, I confirmed that they employ consistent and coherent working methods and that ML has the potential to help augment critical phases in the visual identity process. My findings are further explored via non-participant observation that, in conjunction with the interviews, has led to a primary hypothesis subsequently tested through a within-subject design survey. My findings collectively provide a series of propositions that constitute the basis for a concrete ML implementation proposal. The definition of a replicable conceptual framework that incorporates the shared semantic cognition of design teams into an ML recommendation system constitutes the main contribution to the knowledge offered by my thesis.