Critical thinking is fundamental not only to academic success but also professional competence and informed citizenship. Yet emerging evidence suggests that regular reliance on generative AI (i.e., those digital technologies developed using large language models such as ChatGPT), particularly when performing analytical tasks, can result in lower proficiency during reflection and problem-solving; increased reliance on generative AI appears to be associated with a decline in critical thinking. Such developments pose a challenge to educators and policy-makers. In response, this talk aims to identify and develop a diverse array of strategies for mitigating potential negative effects. To do so, I draw on recent work on ‘epistemic cognition’. There is emerging evidence from this literature that ‘constructivist’ reasoning (i.e., reasoning which holds that knowledge is complex, tentative, actively constructed, and justified via evaluation) helps to improve learning strategies, self-regulation during problem solving, and academic performance. What’s more, such work also reveals an important role for emotions in critical thinking – confusion and anxiety, for instance, have been shown to be positive predictors of critical thinking, leading to deeper processing as well as metacognitive self-regulation. To avoid the negative impacts of increased AI reliance on critical thinking, I suggest that cognitive tasks must be developed in educational and professional contexts with an eye to being intellectually and emotionally challenging, require shifts in perspectives, and focus on building competencies for critical reflection. The upshot is what I call the ‘constructivist framework for AI pedagogy’, an important first step toward a set of conceptual and instructional tools for educators and policy-makers.
Speaker: Luke Kersten