Artificial Intelligence (AI) tools are increasingly being used in education for various purposes. In particular, AI chatbots such as ChatGPT, with their user-friendly interfaces are being explored in education to co-create teaching materials, provide advice and guidance to educators, simulate classroom scenarios, and offer personalized recommendations to students on how to study and approach subjects. With all the enthusiasm for these new opportunities, one should be aware of the risks due to potential biases in the generated content or the responses. These biases can be associated with factors such as gender, race, religion, or political orientation. As a consequence, educators who are using AI chatbots to (co-)create teaching materials need to have the knowledge and the strategies to mitigate such biases. This paper focuses on one particular type of bias, namely gender bias, and on specific disciplines, namely Science, Technology, Engineering and Mathematics (STEM). Gender bias in STEM education is particularly problematic because it may reinforce existing stereotypes about girls and women in STEM and contribute to their underrepresentation in STEM fields. To raise awareness of these risks of gender bias in AI-co-created STEM teaching materials, this paper identifies risks of gender bias by analysing potential usage patterns of AI chatbots by educators when creating teaching materials. An example of such a risk is if the AI chatbot generates educational materials that primarily portray men as STEM professionals and underrepresent women. This would exacerbate the lack of female role models in STEM. Therefore, strategies are developed that educators can apply to mitigate these risks. These strategies will be demonstrated using practical examples. This will allow them to break the vicious cycle of perpetuating stereotypes in STEM education. In addition, these examples demonstrate how AI chatbots can be used to make STEM education more inclusive, which may include co-creating educational materials tailored to individual interests and learning styles.
Read full abstract