Beige by Design
For humanity’s sake, let us be minds,
Not cogs in machines,
Not hands scrolling,
Allowing algorithms as proxies for our thoughts.
With passive queries and vacant eyes,
GPTs complete threads we barely sought to weave.
(With apologies to D H Lawrence)
Minimalist outfits, beige and cream colour palettes, and an air of perfection might not be ubiquitous, but they are widespread, particularly in influencer culture. This beige aesthetic is being claimed as creative output in a lawsuit, between Amazon influencers Sydney Nicole Gifford and Alyssa Sheil. Gifford is alleging copyright infringement on their “carefully curated” beige “clean girl” content. Sheil points out that they are influencing their followers with the help of a spreadsheet sent to all Amazon influencers by Amazon, and so they are all working from the same source material. They cannot help but be similar, close to identical. What does “creativity” mean when there are only so many ways one can model a champagne colored pyjama set against a latte colored wall?
The generative AI and Large Language Models we see being embedded in everyday software and online environments are trained on large bodies of content, frequently scraped from the internet. Without permission from authors (but with the enthusiasm of publishing companies), books and articles are scraped and included into the source material to train these tools. What this means for people who are using genAI and LLMs to generate content (write emails, plan lessons, outline articles, answer essay questions) is that they are reusing what has already been written. And people using the same tools are working with the same source material. What does creativity mean when there are only so many ways to extrude content from the slurry of preexisting work?
What does Beige look like when it is algorithmic? Algorithmic conformity leads us to homogenized cultural outputs, predictable content adhering to the most prevalent norms in whichever training dataset it is trained on, and inevitably the outputs from this beige generator will lead to beige societal outcomes.
How do we get to this beige-on-the-brain state? People have been engaging with this recent article about GAI tools, and the “future of critical thinking,” and many have been struck by the notion of “cognitive offloading,” wherein people who use generative AI to “help them think” end up reducing their active role in thinking, critically or otherwise. From the discussion in the article:
“The study highlighted that users who trust AI tools are more likely to rely on them for decision-making, thus reducing their engagement in critical thinking processes. This trust, fostered by the perceived reliability and convenience of AI tools, promotes a cognitive dependence that diminishes the need for active cognitive effort” (Gerlich 2025)
Human creativity, individual and collective, is sucked into, churned around, and then shoved out again in worse form than before by these machines. There is nothing new, no insight to be gained, no unexpected delights in the world of genAI content.
Does genAI align with capitalism by promoting scalable, reproducible, bland content over unique content? The beige generator not only produces scalable, reproducible and bland content for the masses (or at least what the training dataset suggests are the masses). The AI content will adhere to proven formulas, those most prevalent in the training data, those that yielded maximum engagement, the formula for profit.
On BlueSky in early 2024 Rua M. Williams shared (with permission) a story of one of her research assistants losing themselves in the prose they generated with ChatGPT, out of a worry that their language skills were not adequate (they were! All students need to learn things!). Williams is not the only educator concerned about loss of student voices by using ChatGPT to “clean up” their writing. Jonathan W. Grey discovered that not only was one AI tool “helping” with a student’s grammar, it also “helpfully” deleted all references to Frantz Fanon in the essay! Students told to use ChatGPT to sound “more American” or “more confident” or “more white” (less engaged with Black theorists such as Fanon…?) sound less and less like themselves, and more and more like…something “neutral,” cream, or beige. The neutrality of the GAI algorithms is whiteness, it’s what has been visible and powerful before, it is not the potential for something new, a different way of valuing and representing people.
If you find yourself increasingly surrounded by “students should learn to use generativeAI” apologists, Arthur Perret’s “A Student’s Guide to (Not) Writing with ChatGPT” is a nice antidote, as is this conversation with Chris Gilliard.
We would add our voices to the increasing number of people in education concerned for the preservation of spaces for people to try, to learn, to fail, to be messy. We wish for people to have the freedom and time to figure out what their voices are, could be, what they want to be and sound like, independent of the machinery of algorithmic conformity, that reduces everything to beige.
#Beige #BeigeBullshit #Bland #CleanGirl #Conformity #Culture #GenAI #GenAIRefusal #GenAIResistance #LLM #Whiteness
