Conference Agenda

Since the emergence of Computer Aided Design (CAD) software in 1957, educators have been committed to equipping design engineering students with proficiency in the most widely employed CAD tools in the industry. In the early days, training focused on step by step instructions and understanding CAD theory. Overtime, lecturers adopted more hands-on practice, project-based learning and collaborative learning techniques [1]. The abundance of online resources and tutorials has allowed lecturers, particularly in Higher Education (HE), to focus less on training and more on theory and use of various software and techniques.

Over the last decade, the number, pace and emergence of unique sub-sector software development has grown exponentially, making it difficult for educators to adopt a fixed CAD curriculum.

In the early days, the dominant players in the CAD software arena until the early 2010s were Solidworks, Rhino 3D, and Autodesk Maya, each equipped with its own proprietary rendering engine. However, over the past decade, there has been a proliferation of CAD software alternatives, complemented by independent rendering engines like Keyshot, notably embraced by software applications such as Blender, Gravity sketch, and Sketchup. In particular, Blender and Rhino's Grasshopper add-on have experienced a surge in user-generated custom add-ons, broadening the array of available features and functionalities.

The rapid expansion of Artificial Intelligence (AI) tools over the last two years has introduced a new layer of complexity, as tools such as Midjourney, Stable Diffusion, and Control Net have gained prominence in the industry.

The exponential growth of computational tools available has presented higher education institutions with a perplexing choice regarding which CAD software to teach and the most effective pedagogical methods [2]. This complexity is further compounded by the dynamic nature of software development and the diverse career paths of students, each demanding a distinct skill set for success.

This intricate situation prompts the inquiry into an approach to future-proof and tailor CAD software education to the unique requirements of each student cohort . Perhaps the same tool that advances learning can be used for future-proofing?

Whilst there are examples of successful uses of AI in education in other disciplines, the inquiry using baseline data from undergraduate and postgraduate students at a world leading institution raises additional questions:

• Can we design an AI tool that develops bespoke CAD learning for students?

• Can this learning go beyond simple command instructions and extract higher universal principles of CAD theory?

• Specifically in HE, can AI be used to create individual learning plans based on student’s interests, industry needs and ongoing advances in CAD?

References :

[1] Brink, Kilbrink & Gericke, 2023. Teach to Use CAD or through Using CAD: An Interview Study with Technology Teachers. International journal of technology and design education 33.3: 957–979.

[2] Xie, 2018. Learning and Teaching Engineering Design through Modeling and Simulation on a CAD Platform. Computer applications in engineering education 26.4: 824–840.

[3] Ye, Peng, Chen & Cai, 2004. Today's students, tomorrow's engineers: an industrial perspective on CAD education, Computer-Aided Design, Volume 36, Issue 14, 2004, 1451-1460.

Supervising a PhD thesis implies guiding a student through the mastery of skills and competences essential for a PhD student. Upon completion of a PhD thesis, PhD student should become an autonomous researcher capable of independent research problem-solving and thinking. To do so, they should possess a wide range of abilities to proficiently employ general research methods and tools for collecting, analysing, visualising, and interpreting data. They should also be apt to communicate their research outputs through high-quality papers, presentations, etc. The supervisor should heavily support this process, by being responsible for providing domain knowledge and expertise and guiding students through research/professional development initiatives.

In recent years, academic design research has experienced a paradigm shift with the emergence of various artificial intelligence (AI) tools like ChatGPT. Consequently, these influences reflect on how various PhD studies are conducted and potentially indicate needs for modifying associated supervision processes. Rather than seeking to prevent PhD students from utilising such tools, supervisors should be equipping them with the knowledge to use these new resources effectively and ethically. However, to start with, supervisors themselves have to be aware of these various opportunities offered by AI research tools and be proficient in the use of AI research tools. Certainly, supervisors should serve as role models by actively engaging in their professional growth. Within that context, the incorporation of AI in the supervision process needs to be carefully explored to reap the benefits of such a modified approach in a transparent and ethical manner.

This paper presents an explorative study of PhD supervision activities influenced by AI, outlining the affected skills and competences of a supervisor. The role of AI support will be examined by analysing the design research activities (literature review, data analysis, data visualisation, research communication, etc.), and recommendations will be provided on their inclusion in the PhD supervision process. For that reason, the paper delves into the specific skills and competences required by supervisors and examines how existing AI tools contribute to developing PhD students and their continuous supervision. Acknowledging the unique context of design research, the paper underlines the contextualisation of using these AI support tools for the purpose of further improving design studies.

Emphasising an ethical approach, the paper suggests that AI tools should not replace the PhD student's tasks but rather serve as support, fostering the evolution of their research skills. These findings could be used for integrating AI tools in planning design research methodologies in a more structured and systematic way. By laying out individual activities and related AI-support methods and tools, the role of supervisors extends beyond teaching students to use them; they must also prepare PhD students to master forthcoming AI tools that will soon become integral to the design research landscape.

This research is based on a course developed as a model for design and AI that explores the use of AI in the design process, its shortcomings, and its strength as a design tool. Much of the work generated in class by students were visual communication prototypes but lessons learned can be applied to other disciplines within design. Another goal for the course was to produce a work pipeline for the design process which greatly shortened the production of the concept development stage and allowed for longer periods of evaluation of the content. Interestingly students were more critical of the results due to their assumed role as creative directors rather than production designers. The AI image generators developed concepts akin to dutiful employees who were given direction via prompts. Students responded by shaping the prompts and building on the output by seeding the generator with their results. The rapport between student and generator was immediate, shifting toward clear communication in writing prompts and a much greater focus on ideas that were unexpected, and unique as well as those in line with the student’s initial vision. A model for mapping the process based on the double diamond model, from the British Design Council was reimagined to include innovative processes and user testing to form three stages of divergence and convergence. Thoughtful discussions concerning the design process were particularly insightful challenging students who were familiar with current practices and those who were not but could leverage AI to write a design brief, craft innovative prompts, and critique potential solutions that may have otherwise not been explored due to time or effort requirements.

The course was open to design students and masters students who were attempting to qualify for design degree programs coming from non-design disciplines. Course projects were developed around each of the three stages of the process culminating in a comprehensive project that used the entire process pipeline. In some cases, prototype testing was done which caught early problems and misconceptions about the intended audience. All the students created two concepts and used A/B testing methods to gather user feedback. Among students, the use of AI generators leveled their skills to present sketches or create high-level illustrations of their concepts. Students reported difficulty in achieving the results they envisioned until they developed phrasing strategies that worked but also saw results of audience feedback that responded to their desired communication despite a different representation of their idea. It begs the question, is it communicating the concept or developing the form that determines the success of a design?

This paper presents a scrutinizing attempt to design an artificial design tutor (ADT) that can specifically support development task throughout the phases of product development. In the format of a conceptual paper, it is arguable important to consider how artificial intelligence can support task related aspects, and more complex process-oriented design processes, not as an immediate substitute but as a supplement. With the purpose to present the founding principles of an ADT, the full paper adds insights through a series of interviews with academics and professionals working in the field of AI. Today, we witness how advanced algorithms are used to target and propagate specified suggestions given the stage of design, and anticipate actions needed to validate and secure safe progression. The ADT is designed based on generative AI protocols and follows the escalating trend of utilizing more and more areas with AI tools to facilitate and improve existing processes. From Newsweek magazine alone, it has been stated that an astonishing numerous fully functioning AI apps released every week, and with a growth rate of 38% in 2023. The AI components in an ADT can contribute to improved decision-making processes, where machine learning algorithms may be used to improve the ADT’s ability to recognize and capture user preferences, emerging design trends, and successful design strategies. Consequently, given the range of scope these present, we still have not faced any ADT, which probably is connected to the complexity of the process itself. Given that the design process may clearly shift depending on context, there is still enormous amount of internal data points that could be crunched and analyzed to improve existing processes. Without completely surrender to automatic delivery and design an ADT function as a mediating and cost-efficient step for both companies and institutions. To evaluate, pilot programs in engineering and product design courses provide the opportunity to demonstrate its potential to revolutionize the educational landscape by fostering creativity, critical thinking, and practical skills in future engineers and designers. From a teacher point-of-view, adding an element of support like the ADT, can radically shift how human designated design tutors can enrich and support the depth and authenticity of design projects. On the other hand, there might be conflicting suggestions and more selection, probably adding more emphasis on working on validation and process refinement by students. This paper hope to inspire the community of Design Society and E&PDE in particular to further engage in the potential and risks of AI enabled support, and how ADT may influence existing processes.

This abstract introduces the development of a course level data analytics tool which we’ve called ‘the student record’. This tool aims to transition our course team away from a passive, standardised, compliance-centric institutional approach to instead complement this with a responsive, context-specific and user-centred approach to gathering, analysing and presenting student attendance data at course level. The student record is a relatively simple MS Excel-based system which uses a long-standing total quality management approach (statistical process and control - SPC), as a framework for identifying patterns and interpreting data. This framework helps us gain statistically significant insights which are presented on a configurable dashboard showing flags and recommendations. We feel the tool informs and promotes a more dynamic ‘student engagement’ dialogue between staff and students. In effect it facilitates a rolling academic ‘health check’ to help provide support for students, as well as key contextual insights for module teams and course leaders.

One key attribute of the tool is that it exploits the natural language interface of the ‘Analyse Data’ tool in Excel. It is driven by artificial intelligence in way that is similar (but somewhat more limited) than more open systems such as Chat GPT, Bard and others. Importantly it allows staff to ask questions of the data within Excel itself, without having to write complicated formulas and can provide high-level visual summaries, trends, and patterns using automatically created ‘Pivot Tables’. This has empowered staff with data insights that were previously unattainable or excessively time-consuming using institutional systems.

Central to our approach is the system's legacy development - building a long term knowledge-base to facilitate decision-making that is grounded in robust historical records rather than anecdotal observations or longstanding assumptions in order to foster an evidence-based practice.

Ethical considerations are also at the forefront of our system design, where transparency and data privacy are key, and where accessibility for students’ own data is a priority and is encouraged. For example, the presentation to students of their own data forms part of our personal academic tutorial system where students meet with their personal academic tutors three times per year. The intention here is to foster a reflective learning process and continuous professional development while maintaining data security using simple in-house data management systems.

The full paper will provide a more detailed description of the tool and an evaluation of its capabilities, as well as a critical discussion of the key aspects of its development as mentioned above (e.g. AI, data security, ethics).

In recent years, we have quickly entered a new era that has challenged all areas of life. It seems that some old methods are less efficient than modern methods for doing various things. Educational fields should be updated for future generations, specially design education and related specialties. The problem we are facing today is getting the youth used to using the Internet and its countless possibilities. The main goal of this research is to find ways to teach creativity and maintain the human capabilities of designers. The next goal is to know how to use artificial intelligence as a creative teaching assistant. In order to conduct this research, a problem was raised in the class at the beginning of the brainstorming training session, and three groups of students were asked for ideas to solve the problem. After that, answers to solve the same problem were requested from several artificial intelligence platforms. The answers given by the students were compared with the answers given by the available artificial intelligence platforms. The initial findings show that the diversity of students' ideas is more, but order and classification can be seen in the answers provided by artificial intelligence. Also, the familiarity of students with the environment in which the problem was created causes the difference of some answers compared to artificial intelligence unfamiliar with the environment. Considering that this research started a month ago and is still ongoing; Brainstorming sessions will be repeated in other student groups in other regions. After that, the final analysis and conclusion can be presented.