The development and democratization (making the technology accessible and affordable for everyone) of artificial intelligence (AI) is affecting many areas of our lives, including engineering and education. AI has been developed and used for years, but it has gained public attention primarily thanks to chatbot applications based on large language models (LLMs).

Higher education must also reflect on technological changes. It is not enough to regulate the method and extent of AI use in relation to student work, it is also necessary to explore the capabilities of AI across different fields of science and courses. Many fear that AI will make human knowledge, action, and experience unnecessary. For instance, there are widespread misconceptions that artificial intelligence can fully handle coding tasks.

When we think of 3D modeling, we most often think of the visual design environment, which not only allows the construction of 3D solids, but also provides the opportunity for analysis, testing, and simulation. Newer or more detailed custom functions can be added in the form of scripts, but there are also explicitly code-based modeling systems. At the Faculty of Informatics of the University of Debrecen, bachelor students in Computer Science learn the basics of 3D modeling in the form of an optional course. The course focuses on additive manufacturing and presents modeling methods that meet the requirements of 3D printing. Among various design applications, we also deal with code-based modeling related to the students’ field of expertise.

OpenSCAD is a free modeling application with a unique scripting language to support generative design. OpenSCAD does not focus on the artistic aspects of 3D modelling (creating computer-animated movies or organic life-like models), but instead focuses on the CAD - engineering aspects in a non-interactive way. A compiler reads in a program file that describes the object and renders the model from this code, giving full control over the modeling process. Thus, it is easy to modify any step in the modeling process and create designs defined by configurable parameters.

In the context of code-based modeling, the question can easily arise: who will perform the modeling task: the student or the AI? We wanted to examine whether AI currently poses a real threat to our case. We decided to test the performance of two popular AI chatbot tools, ChatGPT and Gemini, which are the most popular among students and are available for free or with a discounted subscription. For testing, we used tasks that were easy for students to understand and could be easily solved based on short text and visual information. We treated the selected chatbots similarly, providing text and visual prompts at different levels.

In my presentation, I would like to present the results generated by the chatbots mentioned above, which show the current capacity of AI. We know that technology is constantly evolving and learning, but user knowledge, control, and critical thinking cannot be ignored.