http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
생성형 인공지능(Generative AI) 기반의 제품-서비스디자인 아이디어 생성 실험을 통한 가능성 탐색
이영현(Younghyeon Yi),연명흠(Myeongheum Yeoun) 한국HCI학회 2024 한국HCI학회 학술대회 Vol.2024 No.1
고도화된 인공지능의 발전에 따라, 인간 고유의 영역으로 여겨졌던 디자인 분야에서도 AI 도구의 활용 범위가 확장되고 있다. 특히, 생성형 AI (ChatGPT-4, Bard 등)는 디자인 프로세스 중 아이디어 생성 단계에서 활용될 수 있으며, 기존 방법의 한계를 극복하고 더 나은 아이디어를 신속하게 생성하는 데 도움이 될 것으로 기대된다. 본 연구에서는 선행연구를 통해 도출된 아이디어 생성 단계에 활용 가능한 프롬프트 가이드라인과 구조를 재구성했으며, 전통적 방식과 비교 실험을 통해 구체적인 인사이트를 제공하였다. ChatGPT 는 고착화된 디자이너의 사고를 확장하는 수단으로 효과적이었으며, 시간 단축 부분에서 분명한 효율성이 있었다. 또한, 수렴 및 의사결정 단계에서 유용하고 신뢰성 있는 도구로 활용될 수 있는 가능성을 확인하였다. 실증적 연구와 다양한 실험을 통해, 향후 AI 와 인간 협업이 디자인 분야에서 더욱 풍부하고 고도화된 결과물을 도출하는 데 기여할 것으로 기대한다.
제품-서비스시스템을 위한 아이디어 생성과정에서 ChatGPT를 활용한 탐색적 실험
이영현(Younghyeon Yi),연명흠(Myeongheum Yeoun) 한국디자인학회 2023 디자인학연구 Vol.36 No.4
Background : Artificial Intelligence(AI) technology is expanding its utilization across various industries, and in the design field, research and development based on AI tools are actively underway. Among them, enerative AI tools, such as ChatGPT-4 and Bard, possess potential applicability in the ideation phase of the design process. They are anticipated to overcome the limitations of conventional methods, facilitating the rapid generation of high-quality ideas. Hence, this study aims to systematically verify and explore the effects of generative AI on ideation. Methods : Two teams, each composed of four designers, were formed to conduct a comparative experiment between the conventional ideation method and the ideation method utilizing generative AI. They were tasked to generate high-quality ideas over a 4-hour span on the topic of “Healthcare Wearable Devices for Generation Z”. The process was observed without intervention. Following the experiment, the feasibility of AI utilization was confirmed through participant FGI(focus group interview) and IDI(indepth interview). Expert evaluations were conducted to assess the creativity of the ideas generated, and insights were obtained through discussions. Results : The method utilizing generative AI produced 6 more ideas than the traditional method, showing an increase of approximately 1.67 times when compared to the conventional method. The quality assessment also showed that the outcomes of the generative AI method were on par with those from the conventional ideation method. Generative AI effectively broadened the confined thinking of designers and clearly displayed efficiency in terms of time-saving. However, there were shortcomings in contextual consistency and structural completeness, making expert validation and convergence essential. Conclusions : In order to achieve optimal outcomes using generative AI, it is imperative to provide clear preliminary information and to employ specific, structured questions and prompts, as well as effective communication skills when interacting with AI. Discernment and insight on the part of the designer, and high-level decision-making are essential. By rigorously evaluating and refining the ideas proposed by AI based on established criteria, we can pave the way for superior solutions and designs. It is anticipated that future collaborations between humans and AI will yield increasingly rich and sophisticated results in the field of design.