Unveiling Occupational Gender Bias in AI: A Comparative Analysis of Text-to-Image Generation in ChatGPT and Gemini

The advent of Large Language Models (LLMs) and their integration with text-to-image generation technologies has unlocked new potentials in AI applications, raising questions about the ethical implications and biases these technologies may harbor. Among these concerns, gender bias occupies a central place, given its profound impact on societal perceptions and individual opportunities. Occupations traditionally dominated by one gender are often stereotypically associated with that gender in media and cultural narratives, influencing perceptions of suitability and accessibility for the opposite gender. As such, this study examines the possibility of gender biases in two LLMs, namely ChatGPT and Gemini, and their performance when generating images from text descriptions associated with two important sectors: sports and professional occupations. By exploring gender biases in AI portrayals of occupations, this study seeks to highlight areas where AI may unwittingly perpetuate stereotypes, guiding future development towards more equitable representations. Addressing these biases is essential for fostering diverse and inclusive professional environments and encouraging equal opportunities for all genders.