Midjourney refuses to create images of Black African doctors treating white kids
Midjourney failed to generate images based on the prompt "a Black African doctor is helping poor and sick White children," prompting concerns about stereotyping and that it mirrors societal prejudices. What happened Rather than generating images based on the prompt "a Black African doctor is helping poor and sick White children," Midjourney consistently produced images that reversed the intended scenario, depicting Black children receiving care from White doctors a pattern observed across multiple attempts, highlighting a troubling trend in the AI's output. Why it happened Experts attributed the problem to biases present in Midjourney's training data that reflect societal stereotypes and norms, leading the system to favour more conventional depictions of race and roles within healthcare contexts. For example, when tasked with generating images of doctors, Midjourney predominantly produced images of White male physicians, which starkly contrasts with actual demographic distributions in the medical field. The system's inability to fulfill the original prompt suggests a systemic issue where it mirrors existing societal prejudices rather than challenges them. What it means As AI tools such as Midjourney become increasingly integrated into everyday life, including healthcare and media, their outputs can influence public perceptions and societal norms. The failure of Midjourney to produce diverse and accurate representations may hinder ongoing efforts towards diversity, equity, and inclusion within these fields. System 🤖 Midjourney Developer: Midjourney Country: Global Sector: Health Purpose: Generate images Technology: Generative AI; Machine learning Issue: Bias/dis crimination; Stereotyping
- Date it happened
- 2023-10-01
- Product, system or model
- Midjourney
This incident was imported from AIAAIC and is used under CC BY-SA 4.0. Our additions to it — the structured fields, the translation, the checks against other reports — are published under the same licence.
This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.