AI OUTPERFORMS 43 SE GRADUATES IN CREATING CAUSAL LOOP DIAGRAMS OF JANIS GROUPTHINK PHENOMENON

This study examines the capacity of advanced artificial intelligence (AI) systems, specifically Mix-ture-of-Experts (MoE) models like Google’s Gemini Advanced, to generate and analyze Causal Loop Diagrams (CLDs) at a level comparable to human experts. We evaluate Gemini Advanced’s perfor-mance by tasking it with constructing a CLD from Irving Janis’s (1982) seminal work on “groupthink.” The resulting CLD was compared against those produced by a cohort of 60 graduate students with 10 to 15 years of industry experience, who completed the same task as part of a graduate-level systems thinking course.