DEVELOPING TEACHERS’ PROMPTING COMPETENCE USING CHATGPT: A COLLECTIVE ANTI-SHORTCUT APPROACH

The use of generative AI (GenAI) such as ChatGPT highlights the need for educators to develop critical competencies in order to harness its gains and limit its weaknesses for learning and productivity.This workshop aims to engage participants in a collective effort to define what makes for good "prompting competence", that is, how instructions to the GenAI should be formulated by the user to generate desirable outcomes.A starting point is that what is desirable from an educational perspective differs notably from that of production; whereas productivity gains with GenAI are quick, human learning is slow and requires time for reflection and adaptation.Recent research suggests that extensive use of GenAI encourages users to take "cognitive shortcuts"favoring fast and optimal solutions over slower and less practical oneswhich negatively affects users' cognitive abilities.We aim to explore effective countermeasures to such risks.In taking a deliberate "anti-shortcut" approach, participants will carry out a routine pedagogical task (grading a short text) using step-wise instructions (prompts) in ChatGPT.First, participants are presented with some concrete examples of effective/ineffective prompts.Second, small teams will "compete" in prompting ChatGPT to perform as well as possible, compared to a set benchmark (a human professional assessment of the text).Once the "winner" is determined, we will collectively "reverse engineer" what made the winning team's prompting successful and reflect upon its implications for developing crucial competencies (e.g.critical thinking, analysis).The workshop is intended primarily for teachers in higher education with elementary experience of ChatGPT.