Implementing GenAI-Supported Learning in Software Engineering and Computer Science Education using Bloom's Taxonomy

Technical Summary: Implementing GenAI-Supported Learning in Software Engineering and Computer Science Education using Bloom's Taxonomy

Problem Statement
The rapid adoption of Generative AI (GenAI) in Software Engineering (SE) and Computer Science (CS) education has created a dichotomy between the potential for learning support and the risks of superficial learning, over-reliance, and academic misconduct. Current institutional responses often rely on restrictive policies or detection-based approaches that focus on compliance rather than pedagogy. The authors argue that these measures address symptoms rather than the underlying challenge: the lack of explicit instructional guidance on how GenAI can be meaningfully integrated to support cognitive development. Without such guidance, GenAI risks accelerating task completion while undermining deep learning, leading to "cognitive debt" where students possess seemingly correct outputs but lack conceptual understanding.

Methodology
This study employs a qualitative, practice-oriented research design to investigate how explicit instructional guidance aligned with Bloom's Taxonomy can support responsible GenAI use. The research was conducted across two distinct educational contexts:

1. Queen's University Belfast (QUB), UK: A final-year undergraduate Software Testing course (CSC3056) with 150–250 students.
2. Azerbaijan Technical University (AzTU), Azerbaijan: A faculty-wide curriculum update across multiple SE and CS undergraduate programs (e.g., Operating Systems, Data Structures, Advanced Programming).

Intervention: The authors designed and implemented a Bloom-aligned GenAI framework. This framework explicitly articulates the appropriate roles for GenAI at different cognitive levels (Remember, Understand, Apply, Analyze, Evaluate, Create). Key design elements include:

• Cognitive Intentionality: Aligning GenAI use with the specific cognitive level of a task.
• Temporal Structuring: Encouraging independent problem exploration before engaging GenAI for reflection.
• Interaction Modality: Shifting prompts from solution generation to critique, comparison, and error analysis.
• Meta-Dialogue: Embedding explicit discussions about GenAI's role as a tutor/mentor rather than a solver.

Data Collection and Analysis:
Data were collected via anonymous open-ended questionnaires (42 students from QUB) and instructor reflections (11 instructors from AzTU), alongside selected learning artifacts. The study utilized thematic analysis, using Bloom's Taxonomy as an analytic lens rather than a measurement instrument. GenAI tools were used transparently to assist with initial code suggestion and theme clustering, with full human oversight for validation.

Key Contributions
The paper makes three primary contributions:

1. A Bloom-Aligned Framework: A pedagogy-driven instructional design framework that operationalizes responsible GenAI use by differentiating appropriate AI roles based on cognitive levels (Section 3).
2. Qualitative Empirical Evidence: An analysis of student and instructor perceptions across two educational settings, demonstrating how explicit guidance influences usage practices (Section 4).
3. Practical Design Implications: Insights for educators seeking alternatives to prohibition or detection-based approaches, specifically focusing on how to frame GenAI as a learning supplement (Section 5).

Results
The findings indicate distinct patterns in how GenAI is perceived and used when guided by Bloom's Taxonomy:

• Cognitive Level Differentiation: Students perceived GenAI as most valuable for higher-order cognitive activities (Analysis and Evaluation), where it supported reasoning, comparison, and reflection. Conversely, GenAI was viewed as less suitable and potentially counterproductive for foundational learning (Remember and Understand), where premature use led to "illusory understanding."
• Shift in Usage Practices: Explicit Bloom-level guidance influenced students to adopt delayed use (attempting tasks independently first) and reflective interaction (using GenAI as a reviewer rather than a solver).
• Intentional Non-Use: The framework legitimized decisions to not use GenAI for specific tasks, particularly those requiring foundational skill acquisition.
• Learning Curve: Both students and instructors reported an initial learning curve. Students found it cognitively demanding to align their usage with Bloom levels initially, while instructors noted increased workload in designing and communicating these guidelines. However, these challenges tended to diminish as expectations stabilized.

Significance and Claims
The paper claims that the educational value of GenAI lies not in unrestricted use or strict prohibition, but in the intentional alignment between cognitive learning goals, instructional guidance, and learner self-regulation.

• Pedagogical Shift: The study demonstrates that Bloom's Taxonomy provides a familiar, scalable, and pedagogy-driven framework for operationalizing responsible GenAI use, shifting the discourse from "policing" AI to "supporting" learning.
• Alternative to Enforcement: The proposed approach offers a practical alternative to enforcement-focused responses, establishing a foundation for future research on cognitively intentional GenAI-supported learning.
• Modest Scope: The authors explicitly state that the study does not claim to establish causal learning gains or definitive performance improvements due to the complexity of real-world educational contexts. Instead, the significance lies in demonstrating how explicit guidance can shape perceived learning support and usage behaviors, offering a structured mechanism to mitigate "cognitive debt" while leveraging GenAI's capabilities.