Bridging AI and Software Engineering: A Novel Approach to Teaching AI-Intensive Systems

AI-intensive systems have become increasingly prevalent in today's fast-paced world, with examples like Uber and Netflix leading the charge. These systems require high concurrency in data access, fast-changing data streams, and rapid analytics. However, to ensure reliability, maintainability, and compliance, these systems must be developed and governed by software engineering (SE) principles. The integration of data analytics and SE offers unique opportunities for innovation, problem-solving, and decision-making processes. This section explores the need for interdisciplinary collaboration between AI and SE engineers to create robust AI-intensive systems.

The 'Engineering of AI-Intensive Systems' course was introduced at Johannes Kepler University to promote collaboration between AI and SE students. The course aims to prepare the next generation of software engineers to build AI-intensive systems effectively. Key learning outcomes include familiarity with the lifecycle stages of systems and SE, understanding of statistical modeling and data management, proficiency in relevant programming languages and frameworks, and the ability to apply AI and SE concepts to solve real-life problems.

The course employs a combination of lecture-based instruction and hands-on project work. Lectures cover topics such as systems engineering using SysML, the AI systems engineering lifecycle, requirements engineering for AI-intensive systems, and design considerations. The course emphasizes the importance of human-centered AI aspects and ethical considerations in system design. Students are introduced to various tools, platforms, and standards relevant to both AI and SE fields.

A core component of the course is the collaborative group projects. Students form interdisciplinary teams comprising both AI and SE members to work on real-life AI-intensive system projects. Four notable projects developed during the course include a gesture control device for people with tenosynovitis, an AI-powered recipe finder app, an image generator for nature-related descriptions, and an AI-powered sales support chatbot. These projects demonstrate the practical application of course concepts and the successful integration of AI and SE principles.

The assessment process is designed to evaluate students' understanding of both AI and SE concepts. It includes iterative project development with regular check-ins, documentation of requirements and design decisions, and a final project showcase. Students are required to demonstrate their grasp of interdisciplinary concepts through a written exam. The assessment approach ensures that all learning outcomes are successfully achieved, with students showing proficiency in applying SE principles to AI applications and vice versa.

The course faced several challenges, including integrating diverse skill sets, managing different tooling and infrastructure preferences, and balancing theory with practice. The interdisciplinary nature of the teams proved beneficial, allowing students to learn from each other and gain exposure to different tools and platforms. The project-based learning approach, along with peer learning and constant feedback, helped address the challenge of balancing theoretical knowledge with practical application.

The 'Engineering of AI-Intensive Systems' course successfully bridges the gap between AI and SE education, preparing students for the challenges of developing modern AI-intensive systems. By fostering interdisciplinary collaboration and focusing on real-world applications, the course equips students with the skills needed in today's technology-driven market. Future iterations of the course may consider incorporating more advanced topics in AI ethics, expanding the range of collaborative projects, and further integrating industry partnerships to enhance the learning experience.