AI learns by absorbing existing images available online, which, in the case of global health, have historically been largely biased and abusive. Yet, AI is seen as a way of generating universal knowledge and products devoid of context and social meaning. This reinforces perceptions of AI-produced Global Health imagery as a ‘neutral’ visual genre, thereby misdirecting attention from the context in which the ‘original’ images emerged, a context in which the current AI system is still embedded.
Global humanitarian aid organization and human right groups have been quick to adopt new photorealistic generative AI tools. Arguments put toward for mobilizing generative AI includes not invading the privacy of people who are suffering, not exploiting victims, or protecting the identity of people who may be harmed.
Below are three examples. Only one is clearly identified as having been created using AI.
“No matter what, AI tools will always rely on what already exists out there. And from my experience, we know what exists out there, is probably about 5% of reality. Just because of the sheer amount of times that we haven;t been able to find what we needed. And that’s online, let alone offline. We’re starting from a point where there’s already a lot of gaps.”
(Global Health Photographer)
