Design Contradictions: Help or Hindrance?

Aron E. Owen,Jonathan C. Roberts

2024-09-04

Abstract:The need for innovative ideas in data visualisation drives us to explore new creative approaches. Combining two or more creative words, particularly those that contradict each other, can positively impact the creative process, sparking novel ideas and designs. As we move towards AI-driven design, an open question arises: do these design contradictions work positively with AI tools? Currently, the answer is no. AI systems, like large language models (LLMs), rely on algorithms that engender similarity, whereas creativity often requires divergence and novelty. This poster initiates a conversation on how to drive AI systems to be more creative and generate new ideas. This research invites us to reconsider traditional design methods and explore new approaches in an AI-driven world. Can we apply the same techniques used in traditional design, like the double diamond model, or do we need new methods for design engineering? How can we quickly design visualisations and craft new ideas with generative AI? This paper seeks to start this critical conversation and offers practical insights into the potential of AI in driving creativity in data visualisation.

Human-Computer Interaction

What problem does this paper attempt to address?

The paper primarily explores how to utilize design contradictions to stimulate creativity in data visualization design and attempts to address the challenges faced by current artificial intelligence (especially large language models) in handling these contradictions. Specifically:

1. **Positive Role of Design Contradictions**: The paper points out that combining two or more contradictory concepts can spark new ideas and promote the development of innovative designs. However, in the shift towards AI-driven design, existing AI tools (such as large language models) often rely on algorithms that generate high similarity, which contradicts the uniqueness and novelty required for creativity.

2. **Challenges in Integrating AI and Creativity**: Currently, AI systems struggle to effectively handle contradictory elements in design, resulting in outputs that are not clear and lack innovation. For example, the concept of a "circular bar chart" is visually appealing but difficult to maintain data comparability and clarity in practical applications.

3. **Exploring New Methods**: Researchers propose the need to re-examine traditional design methods and explore new strategies suitable for the AI era. The paper analyzes specific cases of how AI can generate creative yet flawed data visualizations, sparking discussions on how to improve AI's design capabilities.

4. **Importance of Multidisciplinary Collaboration**: Solving design contradictions requires not only knowledge from the field of visual design but also interdisciplinary research support, including technologies from natural language processing, cognitive load theory, and other areas, to better understand and address complex design challenges.

In summary, this paper aims to promote in-depth academic discussions on AI-assisted creative design and seeks effective ways to balance the contradiction between visual effects and data accuracy.