The AI shortcut?

How do we need to think differently about teaching and learning as a result of AI.

I haven’t posted for a couple of weeks, partly because of time but also because I wanted to finish Ethan Mollick’s fantastic book Co-Intelligence. Reading the book, my head was flooded with thoughts which made me want to put pen so paper but I needed to digest the ideas. AI is a fast-moving subject and I’m sure my thoughts will evolve over time, but I want to set out some thoughts on the impact of generative AI on education.

In order too use AI, in its current state at least, without the pitfalls such as hallucinations and other errors, then some level of expertise is required. This goes beyond AI literacy. It is not just the skill of using AI. Rather, it is about having the knowledge to better prompt the AI and to quality assure the output of the AI.

This requires an education which develops bodies of powerful knowledge (to borrow the concepts of Ausubel and Michael Young) and cognitive ability (whether you call that cognitive tools or mental models, or critical thinking, etc). This in turn requires a knowledge-rich curriculum and pedagogy which activates hard thinking. Sadly, one of the strengths of generative AI in its current form it it allows a shortcut for task completion, but there is no shortcut to developing bodies of knowledge and cognitive skills.

This makes it more important than ever to develop students metacognitive awareness and to explain to students the purpose of tasks and content, so that they understand how to approach the learning and also why they they should avoid the temptation of the AI shortcut. Within this approach, which ‘breaks the fourth wall’ or ‘lifts the lid’ on the black box, so that students understand the why, there is still plenty of scope for appropriate AI usage. Indeed, getting students to used to using AI is a moral obligation: they’ll be living with AI their whole lives. The type of usage is key: we need to identify, and help our students identify, where AI can be a valuable accelerator of learning and to distinguish this usage from unhelpful shortcutting. In other words, the route needs to be same (no shortcuts), but they might be able to go faster and further with the help of AI.

It is also important for teachers to think about the tasks we get students to do. We need to ask: what is the content and/or cognitive skill we’re developing? How are we getting students to think hard? What are we getting students to think hard about? Generative AI, in its current state, does indeed prompt us to think differently about education, but for the moment, that thinking is about being much more deliberate about working out the ‘why’ of how and what we teach, and do even more to share this ‘why’ with students so that they use AI to accelerate learning, with our guidance, rather than trying to use it as a shortcut which can’t succeed.