CHAT-GPT: A CLEVER SEARCH ENGINE OR A CREATIVE DESIGN ASSISTANT FOR STUDENTS AND INDUSTRY?
Year: 2024
Editor: Grierson, Hilary; Bohemia, Erik; Buck, Lyndon
Author: Maclachlan, Ross John Robert; Adams, Richard; Lauro, Veeti; Murray, Michael; Magueijo, Vitor; Flockhart, Gordon; Hasty, William
Series: E&PDE
Institution: University of Strathclyde, United Kingdom
Page(s): 414 - 419
DOI number: 10.35199/EPDE.2024.70
ISBN: 978-1-912254-200
ISSN: 3005-4753
Abstract
Artificial Intelligence (AI) is an enduring driver for design research and practice [1, 2]; the massive potential offset with concern for the future [3]. A new reality has dawned for practice and higher education with advent of Chat- GPT [4]. The Q3 2023 Engineering Designer magazine (IED) reports ‘How Artificial Intelligence is Transforming Engineering Design: Beyond CAD’. Distinct from research agendas for Generative Design [5] and image-based AI [6], the article highlights the ‘world’s first Chat-GPT designed robot’; Lausanne researchers developing design specifications and concepts using a text-only chatbot. In the nascence of Chat-GPT, we want to understand the extent that our industrial networks and students have usefully leveraged text-only AI. This paper reports on 2 complementary surveys on Chat GPT within: the engineering workplace and; Design Engineering HE. 58% of 120 industrial respondents agreed Chat-GPT should be integrated into university courses prompting a second student focussed survey. Some (26%) engineers are using Chat-GPT without declaring to colleagues and with plagiarism policies referencing Chat-GPT, student use is ambiguous. In industry the most likely (42%) application of Chat-GPT was in ‘research’, responses suggesting the tool as a “clever search engine”. This is also a critical application for students, requiring deeper understanding of differences between search engine results and the more succinct and suggestive framing of information by Chat-GPT. 43% of industrial respondents use the tool to ‘start a new task’, 18% to ‘review work’ completed by a human and a small number (7%) admitting to using Chat-GPT output verbatim. Starting and reviewing work seems likely where we will find acceptable advantages for students. Relatively few (18.35%) industrial respondents saw opportunity for ‘creativity’, and ranked ‘efficiency’ (31.19%), work ‘scope’ (25.69%) and ‘quality’ (20.18%) as likely improvements brought by Chat-GPT. Lower ranking of ‘quality’ perhaps relates to common concerns of ‘mistrust/misuse of chat GPT’ (33.94%), ‘diminished human responsibility’ (15.6%) and the lack of concern about AI impact on job availability (1%). Within our industrial network snapshot, practicing engineers are not using Chat-GPT to the systematic ends suggested by the Lausanne project. Early discussions with students have determined that some are using Chat GPT like industry, but more likely using the tool creatively. We expect the full survey to uncover the extent of this allowing publication of findings and implications for project-based learning and teaching in future curriculum. [1] Herbert, S. (1969). THE SCIENCES OF THE ARTIFICIAL. MIT Press.. [2] Gill, A. S. (2023). CHAT GENERATIVE PRETRAINED TRANSFORMER: EXTINCTION OF THE DESIGNER OR RISE OF AN AUGMENTED DESIGNER. Higher Education, 2, 6. From Proceedings of the Design Society, vol 3, ICED 2023: [3] Müller, B. et al, BARRIERS TO THE USE OF ARTIFICIAL INTELLIGENCE IN THE PRODUCT DEVELOPMENT. p. 757-766. [4] Chong, L., & Yang, M., AI VS. HUMAN: THE PUBLIC'S PERCEPTIONS OF THE DESIGN ABILITIES OF ARTIFICIAL INTELLIGENCE. p. 495-504. [5] Thoring, K., et al., THE AUGMENTED DESIGNER: A RESEARCH AGENDA FOR GENERATIVE AI-ENABLED DESIGN. p. 3345-3354. [6] Brisco, R. et. al., EXPLORING THE ROLE OF TEXT-TO-IMAGE AI IN CONCEPT GENERATION. p. 1835-1844.
Keywords: Chat-GPT, text based AI, language based AI