Educator expertise is needed to work out how AI can uplift, not constrain, cognition

Tom Chatfield has written a white paper for Sage on AI and the future of pedagogy and it captures brilliantly all the various bear traps and category errors that crop up in this debate. This is gratifying in itself, but it also does something even more important – reminds us that the various artefacts of higher education pedagogy and assessment are not the thing itself.

As Chatfield points out:

When children at a primary school write stories and draw pictures, the point is not to supply the world with a stream of winsome content. It is to help them become literate, reflective participants in their society. Undergraduates do not write essays or conduct experiments because the world needs more such things. The process is the purpose.

AI use is ubiquitous, not just in education, but in wider society. The effect of that is that what was once a fairly robust and efficient mechanism for determining an individual’s mastery of knowledge and cognitive capabilities, the assessment “product” is no longer so because AI can (to some extent) simulate such mastery.

One of the aspects of the AI debate I have lately found unsatisfactory is that it treats AI as an efficiency tool rather than a tool that should prompt transformation. Thus, there is debate about which aspects of the learning and education process it is legitimate for AI to assist with, such as summarising article insights, or creating bibliographies, on the learner side, and coming up with assessment rubrics, and giving feedback on the educator’s. That debate assumes that the processes of teaching and learning remain generally unchanged, and that it is possible to disaggregate intellectual administrative work from meaningful cognitive engagement.

The offer is that generating efficiencies in busywork will “free up” time for the meaningful bits (nobody wants to say explicitly that it will help educators deliver more for less or busy students produce assessments more efficiently). The risk is that if it turns out that some of the “administrative” bits are actually quite important to the internal construction of, and coming into relationship with, a structured body of knowledge, then students are stuffed.

The response to the risk cannot be to create rules of thumb for different generic learning processes and assessment artefacts, when in fact the answer to what is reasonable and legitimate can only be answered in disciplinary terms, by pedagogic experts who understand how knowledge is produced and engaged with for their subject.

Chatfield offers an alternative, in which educators think deeply about how AI can be integrated into the thinking processes they are inculcating in students through their teaching and learning practice. This requires a lot of metacognition – the ability to talk about the processes of thinking and understand how it works.

His working assumption is not that higher education should seek to police the use of AI in an unwinnable “arms race”, but harness its affordances: the ways it can help education become an even more effective environment for the development of human potential. He suggests that much of what we already know about education technology is relevant here in terms of understanding where technologies can “uplift” or create dependence. And he suggests that assessment needs to become much more of a partnership endeavour in which students are invited to critically reflect on the role of AI in helping them achieve mastery of a subject. The answer therefore to any question about use of AI in education must be, does it help, or hinder, thinking?

Drafting up policies and traffic light systems will help in the short term but they are mechanisms for creating breathing room, not the end point. In the medium term educators need space and resource to develop a critical and thoughtful accommodation with AI and experiment alongside their students with discipline-appropriate ways of evidencing thinking that depend less on output-production.

And in the long term, as Chatfield argues, higher education as a whole will need to develop a double-literacy:

both human knowledge and our knowledge of how humans learn in systems that uplift rather than diminish cognition.