Constructive alignment in a graduate-level project management course: an innovative framework using large language models

Constructive alignment is a learning design approach that plays a pivotal role in higher education by ensuring a coherent link among learning outcomes, instructional strategies, learning activities and assessment methods (Biggs, 1996). It forms the basis for developing courses that maximize learning experiences, deepen subject matter understanding, and amplify transparency by clarifying expectations and assessment criteria for students (Morselli, 2018). This alignment approach promotes active student engagement, nurturing a sense of responsibility and motivation toward achieving educational objectives (Zhang et al., 2022). Through an iterative process of reflection and refinement, constructive alignment enhances the educational model, focusing on student-centered approaches and strategically equipping learners for success in a dynamically competitive global environment (Biggs, 1996; Loughlin et al., 2021; Roßnagel et al., 2021).

While constructive alignment offers potential benefits, some educators find it challenging to integrate into their courses. Wikhamn (2017) underscores a few challenges to incorporating this approach in higher education, such as the time-consuming nature of the approach, especially when it necessitates a comprehensive redesign of courses or entire curricula. The intricacy involved in coordinating learning outcomes with instructional strategies and assessment methods, as well as gaps in training and lack of knowledge in implementing the approach, add to the challenges. These challenges are further pronounced in engineering education as engineering programs often delve deep into technical and design content. Aligning detailed technical and design content with broader learning outcomes can be difficult, as it requires translating these complex concepts into comprehensible, learner-centric objectives (Dym et al., 2005). Therefore, it becomes imperative to identify and assess tools that can support instructors in developing high-quality constructive alignment to their courses.

Large Language Models (LLMs) are advanced artificial intelligence systems designed to understand, generate, and interact with human language at a large scale. These models are trained on vast datasets of text, enabling them to learn a wide range of language patterns, structures, and nuances. LLMs use deep learning techniques, particularly neural networks, to process and produce language in a way that can mimic human-like understanding and responses (Hadi et al., 2023). Some examples of LLMs include ChatGPT (Open AI), Bert (Google), Electra (Google), RoBERTa (Meta), and ERNIE (Baidu). As LLMs continue to permeate various aspects of the education realm, integrating their use into course design holds immense promise (Rahman & Watanobe, 2023). The strengths of LLMs can be integrated into teaching to unlock opportunities that could aid in personalizing learning experiences and tailoring educational content to individual course needs. These systems can be versatile assistants, providing instant access to a vast array of educational resources and aiding in crafting engaging lectures and course materials. By integrating LLMs into course design, educators can optimize their time, innovate instructional strategies, and ultimately contribute to a more interactive and impactful learning environment for students.

This article presents an innovative framework for integrating LLMs, emphasizing ChatGPT for its proficiency in fostering conversational engagement—a process where interactive dialogue systems, such as ChatGPT, engage users in meaningful conversations, enhancing learning through dynamic interaction—and widespread use to support the implementation of constructive alignment in higher education courses. Through an illustrative case study from a graduate-level project management course, we highlight the benefits and challenges of adopting LLMs to transform instructional strategies and bolster consistency in course development. Our research emphasizes how technology can be pivotal in elevating the educational journey, aiding instructors in refining course design, and enriching the student learning experience.

Biggs’ (1996) seminal work on constructive alignment introduces a strategy for designing and delivering high-quality learning in higher education. Constructive alignment combines constructivist learning theory and instructional design literature (Biggs, 1996). On the one hand, constructivism focuses on the notion that the learners’ activities provide opportunities to create meaning (Ertmer & Newby, 1993). On the other hand, instructional design emphasizes that a course's objectives and the targets for assessing student performance must be aligned. Biggs (1996) argues that integrating constructivism and instruction design can happen in three aspects. First, the intended learning outcomes are clearly stated, intertwined with specific course content, and indicate appropriate performances. Second, the instructional strategies immerse students in scenarios that are likely to provoke these anticipated performances. Third, the assessment methods used to evaluate students are based on the targeted performances. The core principles of constructive alignment remain relatively constant; however, refinements, variations, and developments have built upon Biggs’ original work. Boud (2007) extend Biggs’ model not only to encompass consistency of purpose between the proximate elements of programs but also look well beyond the point of graduation to seek alignment with longer-term purposes. Trigwell and Prosser (2014) argue that introducing qualitative differences into any of the three constructive alignment aspects—specifically, intended learning outcomes, instructional strategies, and assessment methods—has been observed to correlate with variations in the quality of student learning. For example, when educators describe that their instructional strategies aim to encourage students to deepen or modify their existing conceptions and to challenge their comprehension, rather than devoting substantial time to presenting information, it tends to result in students being more inclined to adopt deeper approaches to learning. According to Magnusson and Rytzler (2019), constructive alignment’s focus on goal orientation and standardization aligns well with the market-driven ideals prevalent in higher education today.

Several researchers have utilized the constructive alignment approach to improve students’ learning or identify course design flaws. Croy (2018) used constructive alignment to update the presentation rubric to include individual and group performance to align with the course goals, enhancing students’ satisfaction with the course. McCann (2017) implemented constructive alignment in an economics course, where the instructor found that combining formal and informal feedback was essential to promote students’ reflection on theory application and deepen their learning. Lasrado and Kaul (2021) used constructive alignment in a business and quality management course to introduce innovative ways of designing authentic tasks and aligning them with subject learning outcomes. Chan and Lee (2021) utilized constructive alignment to assess holistic attributes in Hong Kong’s engineering education, revealing a reliance on indirect assessment methods. Their study emphasized the difficulty of transitioning to direct assessments without well-designed, standardized marking rubrics. Caution is warranted when applying Chan and Lee’s (2021) findings to institutions with different characteristics; instructors must consider the unique features of each educational context. From prior research, it can be inferred that implementing a more robust constructive alignment approach is associated with enhanced student learning experiences, demonstrating adaptability across various course contexts.

Nonetheless, gaps in practical applications of constructive alignment still exist. A noticeable gap persists in applying constructive alignment principles to course design (Abejuela et al., 2022; Hailikari et al., 2021; Kumar et al., 2022; Maia & dos Santos, 2022; Tobiason, 2022; Zhang et al., 2022). According to the authors cited above, the limitations of applying constructive alignment theory include the time to restructure, organize, and prepare the course and lectures, and to ensure the curriculum meets institutional quantitative reporting and grading requirements. Cain and Babar (2016) and Teater (2011) mention that educators may face resistance from students in changing from a traditional structure to a more formative way of feedback. Ruge et al. (2019) analyzed the challenges instructors may face in implementing a constructive alignment approach in their courses in Australian universities; they identified factors such as lack of institutional support, limited resources, resistance by individual academics to changing pedagogical approaches, and lack of faculty engagement as possible barriers.

Regarding instructional design in interdisciplinary graduate education, Borrego and Cutler (2010) found that constructive alignment between intended learning outcomes, assessment methods, and learning experiences is severely lacking. Many higher education (especially in engineering) courses prioritize technical knowledge and skills, often overlooking the alignment of learning outcomes, instructional strategies, and assessment methods for a deeper theoretical understanding and practical application. This deficiency can impede students’ ability to connect theoretical concepts with real-world engineering challenges, affecting their readiness for the dynamic demands of the profession. Addressing this gap may be time-consuming and challenging for educators. However, leveraging innovative technologies such as ChatGPT can aid instructors in implementing constructive alignment effectively.

The literature on the application of LLMs in education reveals a broad spectrum of uses. Initial reviews have highlighted the nascent exploration of chatbots in educational settings, underscoring the importance of developing effective learning designs and strategies (Hwang & Chang, 2021; Kasneci et al., 2023). From an instructor’s perspective, recent studies have concentrated on leveraging technology to automate student assessments, provide adaptive feedback, and generate teaching materials, all of which demonstrate a shift toward more interactive and responsive educational tools (Sailer et al., 2023; Bernius et al., 2022; Moore et al., 2022; Sarsa et al., 2022; Zhu et al., 2020).

The adaptability of LLMs in creating diverse educational content, such as question–answer pairs and mathematics word problems, underscores their broad applicability across various educational contexts (Qu et al., 2021; Shen et al., 2021; Wang et al., 2021). Tack and Piech (2022) provide an in-depth analysis of conversational agents in educational dialogues, offering insights into their strengths and areas for improvement relative to human interactions. Additionally, Wang et al. (2023) have pioneered an innovative use of LLMs in analyzing qualitative Student Evaluation of Teaching (SET) comments, enriching the interpretation of SET scores.

LLMs have also been used to foster curiosity and question-asking skills among undergraduate students explaining coding concepts in computing education (MacNeil et al., 2022). Additionally, peer learning has been augmented through LLMs, as evidenced by Jia et al. (2021), who employ Bidirectional Encoder Representations from Transformers (BERT) to evaluate peer assessments, promoting a collaborative learning atmosphere. This body of work collectively illustrates the expanding role of LLMs in enhancing educational practices through automation, personalized learning experiences, and support for both instructors and students.

Conversational AI, including LLMs, has proven valuable in language education; it serves various applications such as acting as a conversational partner, addressing language learning anxiety, providing feedback, and offering scaffolds during language learning (Ji et al., 2022; El Shazly, 2021; Jeon, 2021; Lin & Mubarok, 2021). This diverse body of research underscores the transformative potential of LLMs in shaping various facets of the educational landscape. To our knowledge, this is the first peer-reviewed published work that explicitly addresses the integration of LLMs technologies in the context of constructive alignment within a graduate-level project management course design. This study points to an untapped area of research and application where the potential of LLMs, particularly ChatGPT, in aligning learning outcomes, instructional strategies, and assessment methods remains to be explored. The proposed framework for using LLMs aims to assist instructors in implementing a constructive alignment approach in course design and promoting a more comprehensive and application-oriented education for students.

Constructive alignment synergizes intended course learning outcomes, instructional strategies, and assessment methods. The overarching framework for establishing constructive alignment using LLMs such as ChatGPT within higher education is illustrated in Fig. 1. Although there is no strict sequence for developing constructive alignment in a course, this study delineates a four-phase approach: Initially, in the Preliminary Phase, the user should focus on getting information about the course requirements and input it into ChatGPT. In the Learning Outcomes Phase, the framework formulates the intended learning outcomes for the course, which then transitions to designing the Course Assessment Phase. The final phase involves structuring the course lecture schedule and activities; we call this the Course Schedule and Active Learning Strategies Phase. At each phase, there is an opportunity to revisit previous stages and check the information provided by ChatGPT to ensure coherence and consistency across all course components. Furthermore, given the risks associated with the misuse of AI (IEEE, n.d.) as well as equity, diversity, and inclusion (EDI) concerns (Meyer et al., 2023) associated with LLMs, users should be aware that the output from these models can be biased and inaccurate.

The prompts used in ChatGPT are based on the prompt improvement pattern suggested by White et al. (2023). In this pattern, ChatGPT is requested to ask for additional information to enhance the results. The framework proposed in this research necessitates instructor validation, ensuring that every stage is independently valuable and contributes to a cohesive system that enhances learning outcomes. In addition, the critical role of instructor oversight of ChatGPT’s answers is paramount in addressing the challenge of inconsistency and variability inherent in hyper-scale LLMs. This oversight includes continuous validation of the information provided by ChatGPT, ensuring the educational experience remains aligned, responsive, and consistently high quality.

To explore the proposed framework, we employed a case study centred on a course instructed by the lead author, illustrating the use of ChatGPT v.4 to structure the course’s learning outcomes, assessment methods, and teaching activities using a constructive alignment approach. Introduction to Project Management is a graduate-level course designed for Master of Engineering (MEng) students. Typically taken during the first semester of their program, this course draws students from diverse engineering disciplines, including civil, mechanical, software, electrical, and petroleum engineering, providing a rich context for applying and evaluating the framework. Based on the Project Management Body of Knowledge, the course content covers the basic aspects of each of the project’s five main phases: initiation, planning, team formation, control, and close-out.

Transitioning from an online to an in-person modality, the instructor, with previous experience teaching this course, faced the challenge of adapting to the dynamic classroom environment and ensuring the engagement of approximately 360 students split into three classes. The course calendar description is provided in Fig. 2.

The instructor systematically input data about the course into ChatGPT and engaged in conversation with the tool to assess and validate the framework using AI-generated content. The course instructor conducted and compared the results generated by ChatGPT with the previous year’s course materials to evaluate the effectiveness and alignment of the AI-enhanced approach. This comparison involved several key steps:

Exploring constructive alignment in educational frameworks highlights the importance of structured instructional design. The framework suggested in this paper implemented through a case study in the Introduction to Project Management course showcases how ChatGPT model processes align with the constructive alignment proposed by Biggs (1996). The case study shows that this framework offers a potential method for aligning course learning outcomes, assessment methods, and active learning classroom activities with the course schedule. The course schedule is more than 80% similar to the original course, and the assessment methods are aligned with the course schedule and learning outcomes.

Therefore, the proposed framework can be adopted by other instructors to examine the outcomes of LLMs such as ChatGPT in diverse educational contexts. If proven viable, this framework has the potential to decrease the workload for instructors and help them adopt a constructive approach in course design.

While the proposed framework offers promising insights, it is essential to recognize the inherent challenges of LLMs, including occasional lapses in context comprehension and potential inaccuracies, which underline the importance of human intervention and review. The strengths of the LLMs system in this application to course design were balanced with the authors’ nuanced understanding of project management principles and teaching expertise. It is also worth noting that this study centered on a single course, limiting its broader applicability. As LLMs technologies continue to advance, the relevance and implications of these findings will most likely necessitate revisiting and updating.