DEVELOPING HUMAN-LIKE ARTIFICIAL INTELLIGENCE: IDENTIFYING KEY PEDAGOGICAL PERSONALITY TRAITS

With recent advances of Artificial Intelligence (AI) and its application in medical education, understanding how AI can best support learners in achieving educational outcomes becomes essential. In the context of tutoring, when AI augments academic faculty, mimicking faculty personality aspects may positively impact learners' achievement. The aim of this study was to identify and examine what faculty traits to implement in AI agents. A validated instrument, based on the Big Five Inventory (BFI), was used to support participants (n=72) recalling personality factors of familiar, exceptional educators they've encountered. Results from median dimension score analysis displayed a preference for four broad trait dimensions: Conscientiousness (88%), Agreeableness (86%), Extraversion (75%) and Openness (74%). A strong positive preference for Agreeableness and Conscientiousness and negative preference for Neuroticism related to the Digmans' superfactor Alpha which is related to social development. This study supports the need for an affect element in AI driven pedagogical agents. Furthermore, it clarifies the specific preferences of Singapore based medical students. These findings will contribute to the further development of human-like AI research.