Artificial intelligence is shaping everything from your job prospects to your freedom. And right now, the people programming these systems look nothing like the communities most impacted by them.
Your Face Doesn’t Compute
Facial recognition technology fails to accurately identify Black faces 35% more often than white faces. That’s not a bug—it’s a feature of training datasets that overwhelmingly used white faces. When MIT researcher Joy Buolamwini discovered that facial analysis programs couldn’t detect her face until she put on a white mask, she didn’t just find a technical glitch. She exposed a fundamental truth: AI inherits the biases of its creators and the data it’s fed.
This isn’t abstract. These same flawed systems are being used by law enforcement to identify suspects, by employers to screen job candidates, by landlords to evaluate tenants. When the technology fails to see us correctly, the consequences are real: wrongful arrests, denied opportunities, systemic exclusion automated at scale.
Algorithms Don’t Have Opinions—But Their Programmers Do
Here’s what the tech industry doesn’t want you to know: AI isn’t neutral. It’s a mirror reflecting the prejudices, assumptions, and blind spots of the people who build it. When an AI hiring tool penalizes resumes with “women’s” names or historically Black college affiliations, that’s not the machine making independent decisions—that’s historical bias being codified into code.
Natural language processing models trained on internet data learn to associate certain names with criminality, certain accents with lower intelligence, certain zip codes with risk. Machine learning doesn’t question these patterns—it amplifies them, executing discrimination at computational speed.
The Diversity Deficit in AI Development
Only 2.5% of Google’s workforce is Black. At Facebook, it’s 3.9%. In AI research specifically, the numbers are even worse. The people designing the algorithms that will govern our lives are overwhelmingly white, overwhelmingly male, and overwhelmingly disconnected from the communities their systems will impact most.
This isn’t about hurt feelings or representation for representation’s sake. This is about survival. When the people building predictive policing algorithms have never experienced racial profiling, they don’t think to question whether the historical crime data they’re using reflects actual crime rates or biased policing patterns. When the people designing credit scoring systems have never been denied a loan due to redlining, they don’t see how their “objective” metrics perpetuate those same inequities.
What Happens When We’re in the Room
Timnit Gebru, a Black woman AI researcher, was fired from Google after raising concerns about bias in large language models. Her research team was developing ethical AI frameworks that would have forced the industry to reckon with these issues. Her firing sent a clear message: even when diverse voices are in the room, they’re only welcome as long as they don’t challenge the status quo.
But here’s what they can’t stop: researchers like Gebru, Buolamwini, and Rediet Abebe are building alternative frameworks. They’re creating AI ethics organizations run by and for impacted communities. They’re training a new generation of developers who understand that technical excellence must include social responsibility.
The Power to Program Our Future
AI will determine who gets loans, who gets hired, who gets healthcare, who gets surveilled. If we’re not involved in building these systems, we’re subjected to them. That’s not acceptable.
We need Black data scientists questioning the training data. We need Latinx engineers auditing the algorithms. We need Indigenous researchers bringing traditional knowledge into conversation with machine learning. We need queer technologists ensuring that AI doesn’t perpetuate heteronormative assumptions. We need disabled developers making sure accessibility isn’t an afterthought.
Beyond Diversity Statements
The solution isn’t adding a few minority faces to research teams while maintaining the same power structures. It’s fundamentally reimagining who gets to define what AI should do and how it should do it. It’s compensating community members for their expertise, not just extracting their data. It’s open-sourcing models so they can be audited by the people they affect. It’s regulation that holds tech companies accountable when their systems cause harm.
It’s recognizing that the question isn’t “Can AI be unbiased?” but “Whose values are we programming into the machines?”
Our Algorithm Revolution
The future of AI doesn’t have to repeat the past’s mistakes. But that only happens if we’re not just users of these systems—we’re architects, auditors, and activists demanding better.
Every biased algorithm is an opportunity to build a better one. Every exclusionary dataset is a chance to collect more representative data. Every harmful deployment is a reminder of why diverse voices aren’t nice to have—they’re essential.
We’re not just demanding a seat at the table where AI is developed. We’re building our own labs, our own models, our own vision of what technology in service of justice actually looks like.
Because the algorithms are learning. The question is: who’s teaching them?
