Is AI Really Racist, or Are We Prompting It Wrong?
- S B
- Mar 19
- 6 min read
Updated: Jul 11
Published in Towards AI
AI is the tool, but the vision is human.
Is AI Racist? Challenging the Narrative on Bias and Representation
People often ask whether Artificial Intelligence (AI) image generators are inherently biased or racist, as initial image requests frequently produce results that predominantly reflect White or stereotypical features.
I’m a Black woman and an immigrant. I’ve always been a minority in most environments, whether classrooms, research, or professional spaces. What I’ve learned is how to thrive by understanding the reasons behind biases and proactively moving the needle toward solutions.
Did AI generate a White woman the first time I prompted it? Yes. But instead of accepting that as a fixed reality, I learned how to prompt AI effectively, ensuring that it represents the full spectrum of human diversity.
What is a Prompt?
A prompt is the instruction given to AI to generate a response — whether an image, text, or other content. Prompting is the process of crafting those instructions with specificity and intent to guide AI toward a desired result. AI doesn’t “choose” diversity on its own; it responds to the data it has and the clarity of the request.
The cover image of this article exemplifies this, depicting women of different ethnicities, ages, and features such as freckles, distinct hair textures, and varying eye colors. This image showcases the power of intentional prompting.
Misunderstanding AI
The information or data used to teach AI isn’t alive. It’s an object, just like AI itself. Assigning human qualities to AI — such as racism — is a fundamental misunderstanding of what it is and how it operates.
Instead of learning to use AI effectively, it’s easier to attribute human biases to the technology and trigger emotional responses in the people who use it. Often, these emotional responses reflect human projections based on personal experiences and frustrations with societal biases, rather than actual qualities of AI itself. AI is a tool, and it’s up to us as people to master it.
Problems with data quality aren’t unique to AI. Every field faces issues with biased data collection — for instance, if you open a health study, healthy individuals are more likely to respond. This is called The Healthy Subject Bias. What matters most is how we recognize and address these biases. The key is not to label AI as flawed but to recognize and address these biases in how we design and interact with these systems.
Taking Responsibility
AI reflects biases present in its examples used to teach it, yes. But the responsibility — and the power — rests heavily in our hands as creators and educators.
The outcomes we get from AI tools largely depend on how thoughtfully and deliberately we interact with them.
If you engage the AI simply with “woman,” you might get a default image reflecting what’s most prevalent in the examples from which it learned — often a White woman. But if you specify diversity clearly and intentionally, you absolutely can generate beautiful, authentic representations of diverse individuals.
I’ve created numerous examples in my portfolio that demonstrate this clearly. The image above is particularly special — it’s a recreation of a cherished memory from when my daughter was born, highlighting how thoughtfully AI can be guided to reflect deeply personal and diverse moments.
This intentional approach extends beyond racial representation to include body diversity. Thoughtful prompting can produce dignified, authentic portrayals of full-figured women, celebrating different body types rather than reinforcing stereotypes.
The image above demonstrates how AI can beautifully represent musical icons Aretha Franklin and Tina Turner with their actual body types intact — something critics often claim AI systems cannot do effectively. This wasn’t achieved through complex technical modifications, but through clear, intentional prompting that specified authentic representation of these legendary performers. Even in historical AI-generated portraits, we must be mindful of representation.
AI’s Unexpected Choices
Despite careful prompting, AI can make unexpected creative choices. In one image I generated (see below), an additional subject appeared — a mixed-race woman whose features blended those of the other six, dressed in a neutral blazer.
These AI-generated variations didn’t reflect errors. Rather they helped me to refine my own prompts to better capture the diversity I wanted in the final cover image. This shows that even AI, pulling from the all the information it has been given, knows that diversity exists and can be tastefully represented.
Yet, even when AI generates diverse images, human perception shapes how we interpret them. Our biases influence what we accept as representative. This demonstrates that AI doesn’t just reinforce bias — it can, with thoughtful guidance, enhance representation beyond surface-level inclusion.
Overcoming Personal Bias
We generally don’t recognize our own blind spots when discussing the issue of diversity as it pertains to AI.
Take a moment to reflect: When you see an AI-generated image, what assumptions do you make about the people depicted? Are there any patterns in who you perceive as ‘normal’ or ‘typical’?
We may not always realize it, but AI-generated images designed to show diversity often contain flaws that could have been easily corrected with more intentionality. The weaknesses in these images are not necessarily a reflection of the technology itself, but rather of the person creating them.
For example, I’ve seen AI-generated images meant to depict diversity, yet within a group of ten people, the three Black individuals included all look nearly identical. In other cases, images reinforce stereotypes — such as certain ethnicities consistently being placed in leadership roles while others are shown in supporting positions.
What is happening here? The creator likely sees nothing wrong because, to them, the image is already “diverse enough.” But real diversity isn’t just about numbers; it’s about ensuring meaningful representation. This challenge isn’t just about learning how to use AI — it’s about recognizing our own biases and taking personal responsibility to ensure that the people interacting with our content aren’t overlooked or minimized.
The Role of AI Literacy
Rather than dismissing AI or the people using AI as inherently biased, we can choose to focus on AI literacy. AI literacy is the ability to understand, evaluate, and effectively interact with AI systems.
Just as digital literacy is essential for navigating the modern internet, AI literacy is now crucial for using AI effectively.
AI literacy involves:
Understanding how AI models are trained and where biases originate.
Learning to craft precise, intentional prompts that yield diverse and accurate results.
Recognizing that AI outputs are not fixed but adaptable based on user input and refinement.
By shifting the conversation from blame to empowerment, we can help more people use AI as a tool for inclusivity rather than a source of frustration. Education, not condemnation, will move us forward.
A Call to Action
“We cannot solve our problems with the same thinking we used when we created them.”— Albert Einstein
We need to engage with AI with intentionality. Every time we create an image of a human being; we must do so with dignity and respect. We should carefully consider how each person is presented, and when we are unsure, we either research or ask for help. AI is our creation. It is our duty to ensure it reflects the best, not the worst, of us.
Highlighting bias is important — but imagine the impact if prominent critics focused equally on teaching people how to interact with AI intentionally and inclusively.
We must move the needle forward. Raise critical questions about the limitations of technology but also focus on solutions that drive real progress. Every intentional prompt moves us toward a more inclusive AI. Let’s work together to create AI that reflects the best of humanity. The power to shape AI is in our hands, so let’s use it wisely. No system is perfect because no person is perfect.
AI is the tool, but the vision is human.
Join the Conversation
"AI is the tool, but the vision is human." — Sophia B.
👉 For weekly insights on navigating our AI-driven world, subscribe to AI & Me:
Let’s Connect
I’m exploring how generative AI is reshaping storytelling, science, and art — especially for those of us outside traditional creative industries.
About the Author
Sophia Banton works at the intersection of AI strategy, communication, and human impact. With a background in bioinformatics, public health, and data science, she brings a grounded, cross-disciplinary perspective to the adoption of emerging technologies.
