How Can We Know When AI Extends Rather Than Reduces Student Learning?

Presented by: William Cope

Generative AI has brought a serious problem to education. Speaking colloquially, the “cheating” now possible produces an obvious reduction in learning. More generally, “cognitive offloading” is an equally serious problem. This presentation will trace the peculiar problems AI presents for the evaluation of student learning. Retreating into the superficial signs of learning offered by select response tests is not a solution. We still need written assignments by means of which students can demonstrate “complex epistemic performance” or deep learning. Working towards a solution, I will demonstrate CyberScholar, offering learners full AI support while tracking negative and positive effects in their learning.