Artificial intelligence (AI) is a field that, although has been studied extensively, is filled with many opportunities for innovation due to its seemingly infinite amount of applications. The neural network is a form of AI that can analyze images and recognize them for what they are. A similar form of AI known as GPT-3 is able to use any length of text to formulate coherent sentences and paragraphs. For our project, we tested the idea of combining image recognition AI with GPT-3 to improve the accessibility of websites. Within a website, there are many cases where the images do not have alternate text to aid those who struggle to see or interpret the image. We coded an image recognition AI to recognize an image and provide a one-word caption as its output. We transferred this one-word caption into the GPT-3 to have this form of AI expand the caption into a more comprehensive phrase to obtain a complete alternative text for the image.

Full Poster: Website Accessibility Achieved Through Combining AI Fields