Bending the Iron Triangle in Education: Why AI-Supported Assessment Could Evolve Learning Journeys for the Better

In education, we constantly seek ways to enhance student and workplace learner engagement while improving learning outcomes. Integrating artificial intelligence (AI) into assessment practices offers immense potential, but there are also challenges and considerations that we must carefully navigate.

During the recent (August 6 and 7, 2024) AI in Education Conference held in Singapore, I had the privilege of delivering a keynote address on this very topic. In this article, I aim to share a few key insights from that talk, which are meant to help educators like you understand this evolving landscape of AI in assessment and how it can be leveraged to benefit both instructors and learners.

It is important to remember that integrating AI into assessment is not just about adopting new tools—it's about rethinking how we approach the very process of assessment itself. It is about making our assessments more reflective of our students' diverse needs and more aligned with the skills they will need in the future.

Why AI in Assessment Matters

The conversation about changing our approach to assessment has been ongoing for at least two decades. However, the need to revisit how and why we assess students is more pressing than ever, particularly as AI becomes increasingly capable of performing tasks we have traditionally reserved for human educators.

AI's potential lies not in replacing educators but in augmenting our ability to assess and support student learning more effectively. I’m certain that by this point in time, you have seen at least some ways that AI can help us with various administrative tasks and even student engagement and learning. But have you considered how AI can help us navigate beyond traditional over-reliance on summative assessments—those high-stakes exams that often induce extreme anxiety in students of all ages—towards more formative and (very importantly) integrative approaches so that we are in a better position to offer timely, personalised feedback.

The Limitations of Traditional Assessment

Traditional assessment methods, particularly summative assessments, are deeply ingrained in our education systems (both in schools and in the workplace), going back to the beginnings of mass education that emerged around the time of the first industrial revolution. The modern ideas related to summative assessment and formative assessment emerged in the 1960s and can be attributed to the work of the academic philosopher Michael Scriven (who died only last year at the age of 95). However, summative assessments, designed to evaluate learning at the end of a course or unit–to sum up the learning–can be limiting. They provide a snapshot of what a student has learned but often fail to capture the nuances of their learning journey. By design, summative assessments are a lagging indicator of student learning outcomes, and by definition, since they are a lagging indicator, we have lost our opportunity to intervene if students struggle to understand and apply key concepts.

On the other hand, formative assessments, as originally imagined by Scriven and as they have evolved, have long offered a way to monitor student progress throughout the learning process. However, they, too, have limitations, particularly in terms of scalability and the ability to provide immediate, actionable feedback. This is where AI can (potentially) play a transformative role. Unlike summative assessments, these formative assessments provide the opportunity–as a leading indicator of learning outcomes–to discern how students might perform in high-stakes assessments. That is if we have enough of these formative assessment data points to draw an informed conclusion. Herein lies a paradox with formative assessments: the time we can devote to more formative assessments is inversely related to the desire to have more of these learning opportunities.

Bending the Iron Triangle with AI

In many aspects of life, but particularly in the area of decision-making and project management, there is a concept known as The Iron Triangle, which refers to the tricky balancing act between Time, Cost, and Quality. This concept can also be applied to assessment. Traditionally, it has been challenging to achieve all three aspects simultaneously: large amounts of timely feedback (Time) that doesn’t break the bank in terms of people or technology expenses (Cost) while providing depth and richness in feedback that is clear and actionable for students (Quality).

AI offers some immediate opportunities to "bend" The Iron Triangle–low-hanging fruit, if you will. By automating certain aspects of formative assessments, AI has the potential to reduce the time required for grading and feedback, potentially lowering costs without compromising—and, in some cases, even enhancing—quality. But this only addresses one relatively narrow aspect of the problem/opportunity.

Integrative Assessment: A New Paradigm

One of the most promising applications of AI in education is what I have started referring to as "integrative assessment." Integrative assessment as a concept is not new, but the way I describe it and the way I implement it seem to be a new type of assessment better suited to AI-empowered teaching and learning.

Unlike summative or even traditional formative assessments, integrative assessments can be embedded into every step of the learning journey. They are designed to provide ongoing, real-time feedback that is both personalised and responsive to students' individual needs.

For example, imagine a geography lesson on climate change. Students could use AI tools to analyse data, receive instant feedback on their proposals, and even assess their presentations using AI-powered rubrics. This approach supports continuous learning and allows students to demonstrate understanding in diverse ways that cater to their unique learning styles.

If we have traditionally thought of summative assessments as “high stakes” and formative assessments as “low stakes,” then what about integrative assessments? Perhaps we can consider integrative assessments as having “variable stakes” or “balanced stakes”. Dr Edwin Lim , who served as our “Theme Weaver” at the AI in Education Conference, reminded me of another taxonomy when he articulated the opportunity that integrative assessment represents in this way:

• Summative Assessment = Assessment OF Learning
• Formative Assessment = Assessment FOR Learning
• Integrative Assessment = Assessment AS learning

Addressing Inclusivity with AI

Inclusivity is another critical area where AI can significantly impact assessment. Traditional assessments often fail to accommodate the diverse needs of students, particularly those who are neurodiverse. AI can help by providing alternative means of assessment that align with different learning styles and abilities, ensuring that all students have an equitable opportunity to demonstrate their understanding.

For instance, the proper application of integrative assessment can support differentiated instruction and authentic assessments tailored to each student's pace and learning preferences. This fosters greater engagement and has also been shown to lead to better learning outcomes, particularly for students with special needs.

The Role of Authentic Assessment

Authentic assessment, where students are asked to apply their knowledge in real-world scenarios, is increasingly important. Use of generative AI tools by students and learners is fuelling the conversation about where and how to implement more authentic assessments that align with application and the transfer of learning rather than a surface-level understanding. For example, using AI can facilitate authentic assessment by allowing us to create simulations, situations, and scenarios that will enable students to demonstrate their skills in a practical context. Whether via text, video, or augmented reality, the ways in which assessment can potentially occur could dramatically increase in an AI-mediated assessment environment.

A real-world example comes from medical education, where AI-powered virtual patients can be used to simulate patient engagement. These simulations provide pre-clinical students with a safe, judgment-free environment to practice their skills, receive feedback, and refine their approach before interacting with actual patients. This example shows a practical way that AI can help to “bend” The Iron Triangle: more simulated patient engagements (typically very expensive when human actors are used to play these roles) that are scalable since students are engaging an AI patient rather than an actual human. The critical facet that must be validated is the depth, realism, and quality of engagement and the subsequent feedback that an AI can provide (in this example) to these pre-clinical medical students.

Moving Forward with AI in Assessment

It seems clear that AI can potentially change how we assess student learning. However, this potential should be approached with a mix of curiosity and caution. With this mix of curiosity and caution, I advise educators to proactively explore how AI might be integrated into their assessment practices and to look for ways to enhance learning while maintaining the highest standards of quality, inclusivity, and ethics. Done well, we can create a more equitable and effective learning environment for all students.

?

?