As our team continues to navigate the exciting world of Ai art, we couldn’t help but notice some patterns that correlate to stereotypes existing in everyday media. We took a look into how AI generates images professionals in different occupations using two generators: Adobe Firefly and Midjourney v5.
Although a little less photo realistic at this point in time, we found that Firefly does a slightly better job at straying away from typical stereotypes in most cases. This could be partially due to Firefly only pulling media from their own stock image library. Midjourney represents more of what exists in present day media as a whole. Considering both generators produce 4 images per prompt given, it is interesting to see the lack of variety and diversity within some of the results.
As AI continues to expand its knowledge, it is vital to keep this conversation going. Being mindful of exactly what users are looking to generate, especially when it comes to people, will go a long way in continuing inclusive marketing and helping AI generators develop in a positive direction.