Promoting Women's Participation in STEM through a Digital Tool Based on Large Language Models (LLM)

Despite the growing demand for STEM education (Science, Technology, Engineering, and Mathematics), a gender gap persists, influenced by stereotypes, lack of confidence in mathematical abilities, and inadequate vocational guidance. Practical strategies to increase women's participation in STEM include technological interventions. A vocational guidance agent based on Large Language Models (LLMs), specifically designed for women can enhance STEM education by providing personalized learning and targeted guidance, thus fostering greater female participation in these fields. However, LLMs may perpetuate biases due to their training on human-generated texts. To address this issue, we propose the design of a digital tool based on LLMs, exploring personalization techniques and improvements in these systems. The agent is envisioned to incorporate Growth Mindset Supportive Language (GMSL) to foster a growth mindset, challenge gender stereotypes, and support women's STEM identity.