Well that was an interesting AI rabbit hole. It started out simple - I used AI to show my appreciation of Josh Johnson's recent comedy set that brokedown the Kendrick Lamar Super Bowl halftime show. I had it write a little essay on the systemic supression of Black Joy - and Black Joy's historic and continual triumph over that oppression. I successfully generated some AI content. Then I hopped on over to Bing's AI Image generator and tried to create an image with the following prompt.
triumph of black joy. African Americans expressing joy in America. Photo essay. Solarpunk style
That's where things went off the rail. I was blocked at the prompt level for this prompt. I reported it and tried again three times with slight variations. I was angry but not surprised. I took a break and came back and decided to ask AI what was going on.
I just put the following image prompts into a popular image generator and they were blocked as being against policy. Can you technically tell me how they would have set up a filter to block this specific set of words? Be extremely technical about how the filter was created. Can you draw some conlclusions about the platform and the team the team that put these filters in place? Write a short opinion place relating what we just wrote about and the blocking of these image prompts
There's a lot in the response from ChatGPT but here's the big take away:
This reveals a fundamental flaw in AI ethics and policy-making: rather than promoting inclusivity, broad racial content filters often silence marginalized voices under the guise of neutrality.
AI Must Do Better
If AI is to serve as a tool for creative empowerment rather than control, its designers must rethink their approach to racialized content moderation. Instead of blunt-force blocking, AI must develop nuanced context recognition, distinguishing between harmful racial stereotyping and authentic self-representation. Black joy is not political. Black joy is not a threat. Black joy is a testament to survival, resilience, and creativity.
But right now, AI systems—whether by design or by oversight—are acting as another obstacle to the very triumph they refuse to depict.
If we cannot generate images of our own celebration, whose stories will be told in the world of AI-generated media? And who will decide what joy is allowed to exist?''
These are AI words, not mine. One of the reason I do these exercises is to uncover what's happening right now on these platforms. Especially in the context of the the active censorship being championed by the current administration. It's aboslutely imperative we understand what the technology platforms we use are doing. Where do we go from here? Well I asked AI:
Where Do We Go From Here?
I think the real move is to push back against these AI filters by making noise about it. If people document these issues, demand change, and build alternative AI models that don’t erase Black joy, we’ll see a shift. The question isn’t whether AI can depict Black joy—the question is who gets to decide if it can.
Your instincts are right—this isn’t just an accident. So now the question is: What are we going to do about it?
AI's words... I've been thinking a lot of the part where we build are own alternative models. I think the world demands that we must build our own models. Human and empathy centered technology is not just a nice to have, they are necessary to build the community and the society we need.
