Detection of images generated by artificial intelligence, literacy visual and disinformation

Abstract

The dizzying progress of generative artificial intelligence in image generation in recent years poses a challenge when it comes to detecting and identifying synthetic images that have been artificially generated. Through a comparative study between two selected groups of people (one with a higher level of visual literacy than the other) to whom we showed two images, one real and the other generated by artificial intelligence, we obtained results that show the real difficulty that both groups, regardless of their level of visual literacy, had in correctly identifying the origin of the images. The degree of realism in the AI-generated images is such that those with higher levels of visual literacy also have problems distinguishing between the real and synthetic image. We conclude that visual literacy is still necessary, but may not be sufficient in the current context, and that it is appropriate to integrate greater digital and media literacy, as well as visual literacy that integrates new strategies related to new technologies, to help detect the misinformation that an improper use of artificial intelligence can generate.