The rapid advancement and adoption of generative AI in creative and professional domains mark a significant evolution in technological interaction, raising pivotal discussions on its integration, usability and broader societal impact. This paper delves into the multifaceted challenges and opportunities presented by generative AI, focusing in particular on text-to-image tools and their integration into daily work practices of artists, designers, architects and researchers. As these creators increasingly incorporate AI's inherent unpredictability and ambiguity into their work, a critical examination of the ethical, legislative and environmental implications becomes imperative. We highlight how the User Experience (UX) and User Interface (UI) aspects of these groundbreaking technologies remain mostly underexplored. UX pioneer Jakob Nielsen has recently warned that professionals in the field must adapt to the fast-paced advancements in AI or become obsolete for an industry that is becoming pivotal in the contemporary creative field. He draws a parallel between the current AI era and the dot-com boom of the 1990s, highlighting how UX professionals were slow to adapt to the Internet, which led to missed opportunities and a lack of user-centred design. Consequently, the usability of generative AI tools is a paramount concern, especially as non-expert users increasingly engage with these systems. This paper explores the usability challenges, specifically the articulation barrier where users struggle to effectively communicate prompts to AI, and the consequent need for more intuitive and accessible interfaces. Through comparative analysis and a heuristic evaluation of four leading models, i.e. Midjourney, Dall.E, Stable Diffusion and Adobe Firefly, which presents a unique case study for comparison due to their divergent UI and UX elements, we aim to provide insights into the current state of AI tools and propose pathways for their responsible and effective use.
Read full abstract