ENGINEERING DESIGN – DOES AI CHANGE THE PATH OF EVOLUTION IN METHODS & TEACHING?
Year: 2024
Editor: Grierson, Hilary; Bohemia, Erik; Buck, Lyndon
Author: Woolman, Tim
Series: E&PDE
Institution: University of Southampton, United Kingdom
Page(s): 675 - 680
DOI number: 10.35199/EPDE.2024.114
ISBN: 978-1-912254-200
ISSN: 3005-4753
Abstract
The appearance of engineering design methods has evolved from drawing boards and blueprints to CAD screens and possibly augmented reality. Has decision making evolved, or is artificial intelligence (AI) unlikely to supplant the disciplines and practices adopted in engineering design teaching? AI speeds and extends the processing of communications to simulate responses from natural language inputs, also creating graphical responses by emulating the widely sampled rules inferred from graphics. Can it go further to synthesize decision making in form and material? Certainly entertainment industries create virtual worlds from combining an understanding of creative intent and multiphysics. Generative design distributes material according to rules for both structural performance and manufacturability. Can distributed computing begin emulating expertise applied to optimise utility, aesthetic and tactile appeal? If so, what shall engineers be able to add, that can be taught? It may be misleading to extrapolate from the past, but some reverse predictions may ring true. Leonardo Da Vinci sketched and today’s engineering designers still sketch – we still (should) teach sketching through practice. Do computers read their own and other’s sketches – not yet. IDEO design novel solutions by bringing creative insights together – not mechanistically but through human dialogue, while walking in the shoes of users and other stakeholders. We still (should) teach requirements capture and consultation throughout iterative development. Does ChatGPT start to enquire about the welfare of users and show understanding of what makes them productive, healthy, happy and occasionally delighted – not yet. Manual skills have been applying the craftsmanship we take for granted at a macro scale since cathedrals were first raised far higher than many grand designs, though weaving micro and nano scale solutions is clearly the preserve of automation. The human feel for manipulation materials is innate, hard to copy by machine learning. Though combinations of manufacturing processes get quicker and more accurate, will they sense their outputs, experiment serindipitously and fail often – not yet. Clearly we already trust simulations to tell us whether combinations of materials and geometry will fly, float and house our children’s children in the conditions we foresee. Some simulations can also experiment at the molecular level and even explore what will grow, around and even within us. However we have not yet learned to implement many of the proven solutions to some of the biggest environmental, economic and social challenges. To equip new generations of engineers, shall we perhaps retain our belief in the tools we trust – a sketchpad, a desire to curiously seek first to understand (before being understood) and a healthy workshop to make, break and tinker. To learn to engineer is to learn to control a very small part of the physical world. Then how sustainable the combined effects become is up to us and our retaining the practice of learning from continually examining our methods and results. AI can accelerate our improvements, though let's tread softly and carefully.
Keywords: methods