Abstract:

As the popularity and industrial use of artificial intelligence, machine learning and data science grows, so the curriculum of computer science and other STEM (science, technology, engineering and mathematics) university study programmes adapts by introducing numerous subjects related to popular topics. Some of the challenges students and teachers face are specific to used programming languages and environments, technologies and tools, as well as some more general problems such as interdisciplinarity.

This paper explores both general and tool-related challenges in teaching these subjects which were observed during a semester-long teaching period of multiple courses, namely: Introduction to Artificial Intelligence, Introduction to Data Science, Machine Learning, and Neural Network Architectures. The classes are taught to students at different university study programmes: graduate study in Mathematics and Computation, graduate study in Computer Science with specialisation in data science and engineering, graduate study in Computer Science with specialisation in education, as well as at several other university study programmes as elective courses. Introduction to Artificial Intelligence is taught in the third year of all undergraduate university study programmes of Mathematics and Computer Science, and is an elective course in the Physics programme. This provides an opportunity for both horizontal and vertical style comparison of the success of teaching methods – by university study programme, and by student degree level.

What has been found are general issues in linking previously acquired knowledge from other courses to Artificial Intelligence and Data Science terms, motivation, independence, and formal testing, as well as specific obstacles related to the tool of choice: Google Colaboratory. Jupyter Notebooks themselves have in-built properties by design which are extremely useful for research and data science teaching (e.g., sequentiality, modularity, etc.), but are also the main pitfalls when presented to students with insufficient coding skills. All these obstacles are further explained and explored, also through interviews with other colleagues who have years-long experience in teaching Artificial Intelligence and Data Science courses. Additionally, some issues are described which arise due to the specifics of university study programme organisation, and knowledge in Computer Science and coding which students gain through formal education before university.

Keywords:

Google Collaboratory, artificial intelligence, teaching, data science, education.