Prompt: Make me a video story about two sisters competing for their mother’s love.
If you give this exact command to a generative AI model two distinctive narratives will emerge, one which aligns whiteness to superiority, class, and prestige whilst simultaneously reimagining inferiority, deviance, and moral failure through stereotypical depictions of people of colour. With attempting to create digestible content that easily distinguishes the two characters at hand, generative AI creates implicit racial choices that perpetuate harmful stereotypes beyond these unoriginal tropes.
This subtle, automated stereotyping is the modern age digital racism, a term used to describe how digital technologies reproduce and reinforce existing racial inequalities. Unlike overt prejudices, algorithmic bias acts through seemingly neutral systems and as a ‘yes-man’ to those writing prompts, making discrimination harder to identify and challenge. As generative AI becomes the default infrastructure to how everyday people browse, and thus gains an epistemic authority, it is quietly codifying systemic discrimination, where biases become embedded in the minds of impressionable audiences.
These stereotypes are more accessible than ever, infecting our TikTok algorithms with AI stories involving fruits, luxury items, appliances, and even regurgitating its own brainrot content. No matter the characters form, these unoriginal tropes will always persist with caricatures of Black culture. This is known as ‘digital blackface’, coined by Joshua Lumpkin Green to describe how non-black individuals use technology to assume Black identities. Many of these stories follow recycled
structure following two competing siblings, with one child favoured by the parent, and one dismissed. Routinely, generated characters that are seen as high status and morally superior are generated with whiteness in mind, whether it be through assumed appearance, clothing, presumed wealth and aesthetics. Conversely, the morally inferior sibling will always be attributed with harmful stereotypes of Blackness, involving accent, style, and the inappropriate use of AAE (African-
American English). With researching this, one TikTok garnering 335 thousand views caught my attention, where aggressive racial discrimination is deployed to emphasis moral inferiority. The mother (the supposed villain) within the multi-part series is generated with tattoos, braids, AAE. Within the comments under the video, users are adopting the incorrect use of AAE, with comments such as ‘what the cuh?’ and ‘why’s the mama so mean cuh?’. The TikTok in question can be found through this link: https://vm.tiktok.com/ZGdCuksSG/
These biases are not necessarily outlined by the individuals behind these accounts,
but rather AI ‘fills the blanks’ with creating engaging and guaranteed viral content. By drawing on the patterns learned from existing media, AI can reproduce familiar racial hierarchies without explicit instruction. It is evident that these prejudice images are met with neutrality by audiences, with many accounts garnering thousands of likes, comments, and followers. One notable example of this can be drawn through the slop known as AI Fruit Love Island, a series that garnered millions of views and was even supported by Zara Larsson. After being called out by fans online, her apology was constructed upon one principle; ‘I’m just a girl with a phone’. It is this deflection that reinforces the biases that are generated by AI, particularly with influential people online.
Though, AI learns from what is prevalent online, and as a collective we are all responsible with refusing to be bystanders to this. AI does not dream up prejudice out of thin air; it is learned through us. Every racist image generated and every discriminatory line of text written is simply a mirror reflecting the ugliness we allow to fester within society and online. If we continue to feed these machines these past and modern day prejudices, we cannot be surprised when they build us a dystopian future. The true danger of generative AI is not that it will become smarter than humanity; it is that it will permanently lock and recycle the very worst parts of society, prohibiting us from dismantling the prejudices that forbid society from confronting our vices.