AI With Soul: Ensuring Technology Heals, Not Harms

SOCIAL JUSTICE

Technology is never neutral. The hands that build a tool carry the intentions, the assumptions, and the blind spots of the world that formed them. This has always been true — of the irrigation channels our ancestors built to follow water, of the medicines they extracted from forest and root, of the stories they encoded in stone. And it is true of artificial intelligence.

We are living through a moment that will define the next generation of care — who receives it, who is seen by it, and who is once again left outside of it. AI is not coming. It is already here: guiding clinical recommendations, triaging emergency rooms, predicting which neighborhoods receive public health resources. The question is not whether AI will shape our communities. The question is whether our communities will shape AI.

At Ubuntu Village, we begin from a simple premise: I am because we are. This is the heartbeat of Ubuntu philosophy — not the individual as the primary unit of existence, but the community. Wellness is not a personal achievement. It is collective inheritance.

We bring that lens to every technology conversation. When AI enters healthcare, we do not ask first, “How efficient will this make the system?” We ask, “Whose knowledge is built into this system? Whose bodies trained its data? Whose values govern its decisions?” Transparency, accountability, and community voice must be core to any AI system that claims to serve people — and especially those our systems have historically failed.

Our work spans Kenya, Uganda, and Nigeria. We have watched technology arrive in our communities as something done to them rather than with them—telemedicine apps designed without community input, AI diagnostic tools trained on data sets that excluded Black and Brown bodies, digital health platforms deployed in communities without the infrastructure to support them. And now: AI data centers being sited in communities of color, consuming water and energy resources while the community bears the environmental cost.

Our ancestors were master diagnosticians. They read the body through pulse and color, through the quality of breath and the language of pain. They built knowledge systems across generations—passed through practice, through apprenticeship, and through the body of the community itself. That knowledge was not primitive. It was precise.

AI, at its best, is an extension of that same impulse: to know the body more completely, to catch what the eye misses, to remember what individual memory cannot hold. When designed with equity at its foundation, AI can help clinicians detect illness earlier — particularly in communities where late-stage diagnoses are a function of access, not biology. It can translate complex medical language into the words patients actually use. It can map gaps in care, identify where resources are thinnest, and amplify what community health workers already know about their people.

With purposeful and accountable design, AI can have a transformative impact — not by replacing the healer’s touch, but by extending its reach.

But a tool built from incomplete knowledge will always cut in the wrong direction.

When AI systems are trained on decades of health data collected by institutions that excluded, underdiagnosed, and experimented on Black, Brown, and Indigenous people, they learn the patterns of that exclusion. They do not correct for it. They automate it — and then optimize it at speed.

The evidence is not theoretical. AI-enabled pulse oximeters have been shown to give inaccurate oxygen readings for patients with darker skin, leading to inadequate care — a life-or-death failure embedded in a device sold as precision medicine. AI models have required patients of color to present with more severe symptoms to receive the same level of care as their white counterparts. The algorithm did not intend to harm. But intention does not determine impact.

This is what we mean when we say colonialism does not only live in history. It lives in data sets. It lives in the variables that were never collected, the communities that were never surveyed, the bodies that science treated as aberrations from a white male norm. When we build AI from that foundation, we are not advancing medicine. We are digitizing dispossession.

Privacy is also at stake. AI systems can magnify risks including those related to privacy and security. In communities already over-surveilled — communities that have learned to distrust institutions with their data because institutions have used that data against them — the promise of “smart health” can feel like a threat dressed in good intentions.

Ubuntu Village does not believe in abandoning technology. We believe in insisting that technology serve life—every life, not just the ones the market finds profitable.

First, transparency. Communities must be able to understand how AI makes decisions that affect their health, their children, and their futures. A black box that cannot explain itself is a box that cannot be held accountable. The National AI Risk Management Framework emphasizes safety, security, and fairness in AI applications — but frameworks are only as powerful as the will to enforce them.

Second, community co-creation. Not consultation. Not representation as an afterthought. Co-creation — where the people most affected by a technology are present at the beginning, not invited to react at the end. Patient-centered approaches must guide AI design and distribution. This is not just good ethics. It is good science. Tools built with community produce results that actually work for that community.

Third, fairness and ongoing accountability. Bias in AI is not a launch-time problem that gets fixed and forgotten. It is a living problem that requires continuous testing, correction, and community oversight. AI that works well for one group at one moment may drift, fail, or harm as conditions change. The work of equity is never done.

And finally, most essentially: AI must support human wisdom, not supplant it. The healer, the community health worker, the grandmother who knows her grandson’s body — these are not inefficiencies to be optimized away. They are the living infrastructure of community health. Technology that cannot see them cannot serve the communities they hold.

There is one more layer we cannot name without naming it plainly: high-speed broadband access, critical for equitable AI distribution, is still lacking for 2.6 billion people worldwide. The communities that stand to benefit most from AI-assisted care are often the communities that cannot access it — not because they lack need, but because access to the infrastructure of the digital world is itself distributed along the lines of race, class, and geography.

This is not a technical problem. It is a political one. It is a justice one. And it asks us—those who advocate for technology, those who build it, those who fund it—to extend our imagination past the moment of invention and into the landscape of distribution. Who will receive this? Who will be excluded? What does it mean to call something equitable when billions have no door to walk through?

The Ubuntu Village vision for AI is simple and ancient: build from within the community, not from the outside looking in. Bring the knowledge holders into the room. Honor what communities already know about their own health, patterns of illness and healing, and history with institutions. Then let technology serve that knowing — not replace it.

We are not naive about this. The forces shaping AI development are largely corporate, largely concentrated, and largely driven by profit rather than purpose. But communities that organize, that demand accountability, that refuse to be subjects of systems they did not help design — these communities change what is possible.

Ubuntu means “I am because we are.” AI with soul means a technology that knows it does not exist apart from the people it touches. One of the most intimate frontiers of this question is grief — including the rise of AI-powered avatars that can hold real-time conversations in the voice of loved ones who have died, explored in our piece on generative ghosts: AI-powered avatars that bring comfort after loss, and what this means for ancestral connection. That reminds me of what it was made for. That heals, rather than harms, because the people who built it demanded nothing less.

Not receiving our emails? Re-subscribe here.