Opinion piece.
This post is definitely outside my comfort zone with reference to artificial intelligence (AI) and certainly beyond my qualifications in this area. I am writing as an educationalist and as a human being trying to make sense of my own values of education and my feelings about the potential of AI. There is a buzz about AI in education and I suspect (and expect) my thoughts will evolve over time as I learn more. This is a snapshot of my thinking now and a response to some of the framing of AI in education.
AI efficiency
Within the context of personalised learning, the EDUCAUSE Horizon Report 2023 suggested, as I have read and heard many others previously, the potential for AI to relieve educators from administrative and time-consuming burdens:
“Further, many of these tools are designed to offload the most time-consuming elements of teaching, such as writing assessments, providing students with formative feedback, and making minor grammatical corrections. Spending less time on these tasks should give faculty more time to engage with students directly, tackling more challenging pedagogical tasks such as synthesizing and analyzing information and creating new knowledge.”
EDUCAUSE (2023, p.19)
This places AI as an enabler for educator-student contact, higher order facilitated thinking and learning activities, the creation of new knowledge. This promises the maximisation of productive time in education, but I am not convinced this can be achieved if the human element is overlooked. Learners cannot dive straight into higher order thinking, there is a skillset to develop, there are discipline practices to establish, and these take time. There’s a balance needed then between what is truly adminstrative versus what learning and support is dependent on human interaction. The consequences may be different in various educational settings.
There is one aspect here that I think needs addressing and I appreciate that there are varying interpretations of how AI may support assessment processes. However, assessment and feedback should not be a one-way activity. Feedback, particularly formative feedback, is not about correcting learners mistakes. Feedback is about enabling learners to learn from mistakes, to support the appreciation of why, to enable them to approach the next problem with greater understanding. This requires a level of responsiveness not just in the moment, but responsiveness to learning design also as courses are iterated. Educators have much to gain from the dialogue of feedback, by identifying student misconceptions, challenging ideas and finding new representations of subject matter. This dialogue can utilise data, but should not be assumptive about the reasons behind it. Learning is two-way between educators and learners. Human interaction is part of the process, though I admit it does not need to be the whole process all the time.
AI disruption
With reference to generative AI specifically, the Horizon Report presented the call to rethink assessment models and move away from recall and knowledge-based assessment. Recent examples from generative AI suggest mimickry of higher order evaluation and synthesis as well, so it’s not inconceivable that assessment models have to go further.
“As technology advances in ways that make basic information retrieval available to us at all times, assessing students’ ability to memorize and repeat information is arguably obsolete. Generative AI presents an opportunity for educators to challenge mainstream assessment practices and shift their focus on students’ abilities to practice higher order skills such as analysis and evaluation.”
EDUCAUSE (2023, p.21)
With generative AI challenging the notion of learning evidenced through outputs such as written assessments and even creative works, I am wondering where this leaves formal education. Perhaps one avenue to explore is not outputs, but actions. This is not an easy measure to scale within formal settings, but is, I’m somewhat asserting, authentic assessment. From learning, how learners individually contextualise and build their understanding is then demonstrated through the decisions and actions they then take as a result of that learning. The challenge is that what is learnt may not have authentic demonstration in a learner’s application of that learning through action until much later, if ever, in that learner’s life. Consider that formal curricula may teach a broad range of topics that then bear relevant application within specialist careers and specific projects. Hence, the necessity for constructed assessment activities.
AI shifting education into contexts
There is another way to look at the role of formal education alongside ‘real world’ application that may lead closer to maximising productive education. Van Borstel (1992) provided a rather utilitarian view of productive education by grounding formal schooling activities towards economic and social outputs. I don’t fully subscribe to that notion, as learning for personal development and enjoyment is as much a part of educational experiences, but how productivity influences policy may also influence the role of AI. AI and productivity are often spoken in tandem and there is an implicit, if perhaps unqualified, assumption to that effect. However, perhaps productivity comes as a consequence of AI disrupting educational approaches, rather than productivity through efficiency. If generative AI makes constructed scenarios of assessment redundant, then this opens the possibility of placing the educational process within the context of direct ‘real world’ application as more appropriate. In a way, this is similar to degree apprenticeship models, and indeed a number of vocational programmes, where academic study and workplace learning are integrated. In these cases learning can be directly applied to real challenges, there is reduced abstraction and immediate demonstration of learning through action. This changes assessment of learning. What a learner, as an individual, does, who they interact with, what they produce and what they solve, with or without AI as an aid, can be measured. It’s time intensive, but during or after a programme of study this immediate selection of knowledge and understanding is combined with personal attitudes to tackle authentic problems. Synoptic assessment approaches are possibly the closest to this paradigm. The added aspect here is bringing the learner into a discipline, so that learning is more than just the abstract processing of facts and techniques. Feedback though would be more complex and nuanced, but could be distributed between formal educator, self-assessment and peer- or colleague-assessment.
AI, education and the human element
In possibly the purest sense of personalised learning, it’s important to consider that what is taught is not all learnt, and what is learnt is not all applied. As subjects are taught, the decisions learners make of relevance of new material, based on comprehension of new knowledge and foreshadowing its potential benefit, are in juxtaposition with formal curriculum design that purposefully scaffold and progress learning over time. Yet, this where human connection comes in. The narrative and the stories of educators to bring learners into the discipline, the art of representation, the building of relationships between subject matter, these are all human expressions of a field of study. These narratives are our ways of knowing, our ways of expressing our understanding. This connection between educator and student aims to bring relevance to what may otherwise be perceived as irrelevant.
In this human endeavour of education, educators are empowering learners to make decisions of learning based on experience shared, empowering learners to bring their own meaning. That’s what makes the difference between learning from an anonymous text or an AI bot, and learning from another person. Connection is key and that’s why my personal view is that we have to ensure connection remains a priority in the design of educational experiences.
It’s rather apt then that the summary of the Horizon Report described the expert discussions as:
“… seemingly polar ideas: the supplanting of human activity with powerful new technological capabilities, and the need for more humanity at the center of everything we do.”
EDUCAUSE (2023, p.4)
Yet, we don’t have to treat these as polar opposites. There is a reason why students learn at universities and there’s a reason why professional learning takes so many forms in so many places. We utilise learning spaces, access learning materials, generate lived learning experiences and belong to learning communities in ways that vary by individual. Learning is personal, the process of learning is often inefficient and it is frequently complex. These are human qualities that any AI enhanced educational environment should respect and complement.
If AI can replace educators and educational professionals, then we are doing education wrong. If AI can make assessments meaningless, then we are doing education wrong. If AI can make knowledge and knowing redundant, then we are doing education wrong.
References
EDUCAUSE (2023). Horizon Report: Teaching and Learning. EDUCAUSE. Boulder, CO.
Van Borstel, F. (1992). A theoretical framework for productive education. Prospects: Quarterly Review of Education, 22(3), 265-271.
Leave a Reply