These images were generated using 2026 Meta AI image-generation software. While newer AI models have produced significant improvements in image quality, realism, and visual detail, technological progress does not necessarily mean progress in representation. Despite appearing more realistic, these images continue to reproduce longstanding and harmful visual tropes, including patterns in how race, healthcare, and power are portrayed.
As AI-generated imagery becomes increasingly integrated into education, research, media, and public health communication, it is essential to ask: How long will it take for AI systems to recognize and address the biases embedded within their training data?
The images below, mainly from the photography collection of the former Canadian International Development Agency, represent different approaches to visually representing some global humanitarian aid issues.
Answer the questions for each pair of images.
Which image would more likely get you to act to end hunger?
Which image would more likely get you to support health equity and universal healthcare?
Which image do you think better serves refugees?
If not AI then what? Models/actors? Community collaboration?
Share your comments on these questions & the exhibit here:
