AN ANALYSIS OF THE ACCURACY AND BIAS OF A GENERATIVE AI MODEL

Abstract

Artificial Intelligence (AI) has become deeply ingrained in the daily lives of ordinary individuals with tools like ChatGPT, Google Bard, and Amazon Alexa playing integral roles. The increasing reliance on AI prompts questions about its true capability to meet diverse needs. Facial recognition and generation stand out as a significant area where AI is still evolving AI continues to encounter challenges in both the detection and generation of emotions. This study utilizes the popular image generatorMidjourney to examine the efficacy of generational AI. Concerns arise about the potential misuse of these tools to disseminate misleading information. The research aims to assess the accuracy and bias of images generated by one of the most advanced image-generating algorithms by seeking evaluations from human volunteers. Results indicate notable biases in gender, race, age, and physical abilities, raising concerns about the inclusivity of AI-generated content. The researchers conclude with insights into the limitations of current AI models and issues of accuracy when generating emotion-based prompts. Suggestions offuture research are provided, emphasizing the need for larger sample sizes and focused studies on specific emotions and biases.