Abstract:

The rapid integration of generative AI tools into everyday learning has fundamentally shifted student expectations in computer science education. Where traditional AI and data science courses emphasized manual implementation of algorithms, environment setup, and syntax mastery using tools like Google Colaboratory and Jupyter Notebooks, today's students increasingly expect to focus on problem framing, model interpretation, and ethical deployment, using AI to handle boilerplate code.
Building on our prior work identifying key pedagogical challenges in teaching AI subjects, including difficulties linking prior knowledge, maintaining motivation, and navigating tool-specific pitfalls like notebook sequentiality, we conducted a follow-up study comparing instructor intentions with student expectations across multiple courses at the Faculty of Science, University of Split.

Survey and interview data from students enrolled in Introduction to Artificial Intelligence, Machine Learning, and First year project, reveal a growing disconnect: while instructors prioritize foundational concepts and coding competence including algorithmic understanding, students exposed to AI-assisted development workflows value rapid prototyping, numerous prompt-based iteration, and system-level reasoning. Notably, challenges previously attributed to insufficient coding skills are now compounded by mismatched expectations about the role of code itself. Students who successfully engaged with course material were those who received explicit scaffolding on when to code manually, when to delegate to AI, and how to validate AI-generated outputs.

This paper presents these new findings and proposes a responsive pedagogical framework that preserves conceptual rigor while accommodating AI-augmented workflows. We outline practical strategies:
(1) restructuring assignments to assess design thinking alongside implementation,
(2) integrating "prompt literacy" as a core competency, and
(3) adapting assessment rubrics to evaluate reasoning over syntax.

Keywords:

Artificial Intelligence, Generative AI, Student Expectations, Curriculum Design, Assessment Methods.