Data-driven ideas

In the early days of design research - 1960s, a scientific approach to to the design process was promoted as the result of new technical developments such as computers and automation. Although the design science relationship was, and remains controversial, scientific knowledge, and its application is still relevant to understanding how designers think and work (Nigel Cross, 2023). A recent example is GenAI models trained on terabytes of web crawl data from across the internet to create new content. Being data-driven means that GenAI's output is derived from data analysis, which is part of the scientific method. In the data-driven approach, GenAI represents a break from the more traditional view of design as a creative practice supported by artistic, intuitive processes or personal opinions. Yet GenAI does not exclude the designer from the ideation process because GenAI operates on a probabilistic framework, it does not possess true understanding or consciousness like a human. That is, GenAI is not capable of learning, or understanding the concepts underlying its own responses to human prompts, which are input elements for GenAI to generate results. Indeed to achieve human-centered desired outcomes, writing effective prompts is considered a skillful craft. And so, although GenAI evolved from computer science, the technology is guided by human prompts. GenAI, then, reflects human-machine interaction -  a tool that facilitates ideation.

Easy Aideas

Design ideas typically are about incremental change or improvement over something that already exists rather than about something that is truly original or radically different. For incremental innovation, then, generative artificial intelligence has emerged as a powerful ideation tool that facilitates a seemingly endless flow of creative content. But how does GenAI tools like ChatGPT and Copilot compare with human ideation without GenAI? GenAI foremost advantage is it generates huge amount of diverse content - both text and images using pre-existing data from across the internet. Then there is the ease of use of GenAI as it responds to written prompts in a conversational style and in multiple languages - and at great speed. More, as GenAI is going mainstream it has the potential of facilitating problem solving on a global scale. But there are weaknesses with GenAI models.. For example, writing effective prompts may seem simple but rests on prerequisite knowledge, language skills and, yes human imagination. Even so, the user friendliness of GenAI and the creative and original appearance of its output carry the risk of over-reliance on GenAI. Indeed users might find themselves having more ideas than they know what to do with. Also, it is not transparent what data GenAI models are trained on - raising ethical or copyright issues. Or, the self-referential loop of GenAI data might, paradoxically result in more similar output over time narrowing the scope for plurality or novelty. But values, assumptions and biases are embedded in GenAI tools, and so more empirical evidence is needed to fully evaluate the pros and cons of GenAI systems. Yet the appeal of GenAI is overwhelming and by fusing AI-generated ideas with human judgement and refinement, it is fair to say that GenAI is enhancing human creativity, including ideation.