Artificial intelligence did not create racial inequities in America. Humans did. Long before ChatGPT, Claude and other generative AI tools existed, Black children were disproportionately miseducated in underfunded schools. They received outdated textbooks that had been discarded by wealthier, predominantly white schools. Black students were overrepresented in special education and school suspensions, while simultaneously being underrepresented in gifted programs and advanced courses. None of this was technologically engineered. The inequities were not artificial.
This distinction matters as Americans become increasingly concerned about racial biases in artificial intelligence. Algorithms have indeed been found to disadvantage Black people in hiring, healthcare, policing, lending, housing and other domains. But blaming the technology risks obscuring the fact that humans write the code, select the data on which AI systems are trained, approve their use, and decide where and how they are deployed.
We should be equally concerned about who gets access to these technologies. My work has taken me to hundreds of K-12 schools, colleges and universities across the country. I have visited extraordinarily wealthy private schools in which every student has a laptop, as well as underresourced public schools that enroll mostly Black and Latino students, where computing technologies were far less plentiful and current.
The COVID-19 pandemic made these longstanding inequities impossible to ignore. While students in affluent households logged onto remote classes using reliable high-speed internet, many low-income children and college students of color sat in cars outside libraries and businesses to access free Wi-Fi. AI did not create that and other digital divides. But it could significantly widen it.
Credit: Contributed
Credit: Contributed
Universities are creating massive AI and computing institutes, corporations are spending enormous sums developing new technologies, and schools are figuring out how to prepare students for an AI-powered economy. This raises an important racial equity question: Where is all the money going?
We should know how much is being invested in predominantly Black K-12 schools, community colleges that disproportionately enroll Black students and historically Black colleges and universities compared with wealthier, predominantly white institutions. Black students cannot again be the last to receive technologies their white peers have been benefiting from for years.
Previous technological inequities should make policymakers, philanthropists, educators and tech executives especially attentive to this possibility. Equitable access cannot be something we attempt to engineer after a new generation of racial disparities becomes obvious. Governments making AI investments should disclose which schools and communities benefit from them. Philanthropists should deliberately invest in institutions that have historically received less. Tech companies should be expected to show who gets early access to their educational products and workforce development opportunities. Otherwise, we risk treating another predictable racial inequity as an unfortunate technological accident.
There also are racial consequences for the workforce. Some jobs will be eliminated as artificial intelligence continues to evolve. New ones will be created. Black workers cannot be disproportionately concentrated in occupations AI makes obsolete while simultaneously being underrepresented in the professions it creates. Ensuring that Black students develop sophisticated AI literacies in schools, colleges and universities is, therefore, an economic imperative. But employment is only part of the issue.
Black people also have to be among those who create AI technologies. In my work with several Big Tech companies over the years, I have repeatedly noticed the underrepresentation of Black professionals. This matters beyond workforce diversity. When Black people are absent from the processes through which algorithms and technologies are developed, so, too, are perspectives that could help identify and interrupt anti-Black algorithmic biases before they are technologically scaled.
Black Americans cannot leave all of this to schools, tech companies or government. Black parents should ask what their children are being taught about AI and how much their schools are investing in it. Black alumni should ask their HBCUs what resources they need to prepare students for emerging technologies, then help secure those investments. Black churches, historically Black sorority and fraternity alumni chapters, civil rights groups and community organizations should create out-of-school opportunities for young people to develop AI skills.
We also have to show up where consequential decisions are being made. School board meetings, city council hearings, and legislative sessions are places where policies concerning AI will be determined. Black people deserve a say in which technologies are deployed in our schools, workplaces and communities, how they are used and who benefits financially from them. None of this requires every Black person to become a computer scientist. But all of us should understand enough about this technological revolution to recognize when our communities are being excluded, exploited or harmed.
The responsibility cannot belong only to Black Americans. The people and institutions with the greatest power over AI have an even greater obligation to prevent it from reproducing racial inequities. They cannot build systems that disadvantage Black people, distribute their benefits inequitably and then blame the technology for the outcomes. Algorithms are artificial. Racial inequities are not. Structural and systemic racism created them. And humans have sustained them across generations.
When an AI tool produces racially inequitable outcomes, blaming the technology is insufficient. Someone built it. Someone trained it. Someone approved it. Someone decided where and how it would be used.
America has experienced previous technological revolutions inequitably. There is no defensible reason this one should be the same. Artificial intelligence did not invent anti-Blackness. Moving ahead, we ought not train machines to reproduce it.
Georgia native Shaun Harper is provost professor of education, business and public policy at the University of Southern California, where he holds the Clifford and Betty Allen Chair in Urban Leadership. He also is founder of the National AI Equity Lab.
If you have any thoughts about this item, please go to our AJC Get Schooled Facebook page. If you’re interested in writing an op-ed for the AJC’s education page, drop us a note at education@ajc.com.
About the Author
Keep Reading
The Latest
Featured
