Beyond Human in the Loop: Why AI Needs Humanity in the Loop

This past week I had the privilege of moderating the panel “AI Policy for the Public Good” at the 2025 Technology Law and Policy Center Summit at North Carolina Central University School of Law. As I moderated the discussion among Professor Margaret Hu (William and Mary), Professor Keith Robinson (Wake Forest), and Dr. Siobahn Day Grady (North Carolina Central University), the conversation repeatedly circled back to a familiar refrain: we need “humans in the loop.” It’s become almost liturgical in AI ethics discussions, the idea that human oversight ensures responsible artificial intelligence. But as I listened, a realization crystallized for me that I shared with the panelists and attendees in real time: having a human in the loop is necessary, but it is woefully insufficient.

What we actually need is humanity in the loop.

The Flaw in “Human in the Loop”

The problem is deceptively simple: humans can be bad actors. Not all human judgment is ethical judgment. History overflows with examples of humans making decisions that served their interests while devastating others. A human can be in the loop and still perpetrate injustice, extract unfair value, or prioritize profit over people.

We’re already witnessing this dynamic in what Cory Doctorow aptly termed the “enshittification” of digital platforms, and now, of AI itself. Doctorow coined this term—named the American Dialect Society’s 2023 Word of the Year—to describe how platforms initially benefit users, then exploit users to benefit business customers, and finally exploit those business customers to extract all value for themselves. The pattern is predictable: platforms launch with genuine benefits for users, vanquish competitors through these user-friendly features, and then systematically erode value as they pivot toward profit maximization. The human decision-makers in these loops aren’t malfunctioning; they’re functioning exactly as self-interested actors often do.

When we import this same “human in the loop” framework into AI governance without interrogating what kind of humans and what kind of values they embody, we’re building ethical guardrails on sand.

What Does “Humanity in the Loop” Actually Mean?

Humanity is more than biology. It’s a constellation of values, commitments, and practices that call us beyond narrow self-interest. If we’re serious about embedding humanity in the loop, we must grapple seriously with what it means to be truly human: not merely Homo sapiens making decisions, but people exercising our highest capacities for wisdom and moral reasoning.

Humanity in the loop means:

Connection over isolation. It means recognizing that we are fundamentally relational beings. AI systems shaped by humanity in the loop thinking would prioritize strengthening human bonds rather than fracturing them, building bridges rather than echo chambers.

Others-centeredness over self-interest. This requires thinking beyond our own needs and genuinely focusing on the flourishing of others, especially the vulnerable, the marginalized, those whose voices are often excluded from the rooms where AI decisions are made.

Universal benefit over particular advantage. Making the world a better place for all, not just for some. This means resisting the siren call of optimization that benefits shareholders while harming workers, or efficiency that serves the Global North while exploiting the Global South.

Stewardship over extraction. Being responsible caretakers of this planet and all it contains, including the natural resources consumed by AI’s voracious computational appetite and the communities affected by AI deployment.

The Moral Architecture of Humanity

But humanity in the loop requires even more. It encompasses being humane: operating with compassion and recognizing the inherent dignity of every person touched by AI systems.

It means applying the Golden Rule as a design principle: creating AI systems we would want governing our own lives, evaluating our own applications, or making decisions about our own futures.

It demands kindness: not the performative politeness of corporate DEI statements or the sterile political correctness of carefully focus-grouped language, but genuine kindness that costs something and changes how we build.

It requires civility, respect, and more than mere tolerance. Tolerance is not only passive but also othering and distancing. It allows us to keep people at arm’s length, acknowledging their existence without genuinely connecting. Humanity in the loop, by contrast, seeks points of connection rather than points of disconnect. It actively honors the other and builds bridges of understanding. This distinction matters enormously when we’re encoding values into systems that will shape millions of lives.

And perhaps most challenging: it means forgiveness. When mistakes are made (and they will be), humanity in the loop creates space for acknowledgment, learning, and restoration rather than merely defensive legal posturing or blame-shifting.

The Path Forward

Moving from “human in the loop” to “humanity in the loop” isn’t merely semantic. It’s a fundamental reorientation that asks deeper questions:

• Not just “Is there human oversight?” but “What values are guiding that human?”
• Not just “Can we explain this decision?” but “Would we want this decision made about us?”
• Not just “Is this legal?” but “Is this just?”
• Not just “Does this maximize efficiency?” but “Does this honor human dignity?”

This shift requires us to think long and hard about what it means to be human in the fullest sense. It demands that we look unflinchingly at our capacity for both good and ill. And it insists that we build AI governance structures that actively cultivate our better angels rather than simply constraining our worst impulses.

The stakes couldn’t be higher. AI is being woven into the fabric of human life, making decisions about credit, employment, healthcare, criminal justice, and countless other domains. If we settle for merely human oversight without interrogating what humanity requires, we risk encoding and amplifying the very injustices we claim to prevent.

A Deeper Foundation

As we stand at this crossroad, we need more than technical solutions or regulatory frameworks, though both matter. We need a richer moral vocabulary and a willingness to ground AI ethics in substantive visions of human flourishing.

Humanity in the loop is more demanding than human in the loop. It asks more of us. It requires wisdom traditions, moral formation, and ongoing vigilance against our own capacity for self-deception.

But if we’re going to create AI systems worthy of the world we hope to inhabit—systems that serve genuine human flourishing rather than narrow interests—then humanity in the loop isn’t optional.

It’s essential.

What do you think? How might we move from human oversight to genuinely humane AI systems? I welcome your thoughts and pushback.