Generative AI is moving quickly into the ordinary routines of work, learning and communication. It is being framed as a breakthrough in productivity, access and personalised support.
But beneath the excitement sits a harder question. What happens to our ideas of care, obligation and community when more of our relationships are mediated by machines?
That question matters globally. In South Africa, it carries particular weight.
We live in a country where social trust is fragile, inequality is persistent and human institutions are often under strain. We also live with a moral tradition that has long pushed back against narrow individualism. Ubuntu, often expressed as “I am because we are”, reminds us that personhood is not formed in isolation. It is shaped through our obligations to one another.
This is why the rise of humanlike AI matters beyond the technical realm. It is not only changing how we work, search and communicate. It may also begin to reshape how we understand care, attention and presence itself.
That is not just a technical shift. It is a social one.
When machines start sounding human
On the surface, the promise of generative AI appears reasonable enough. A chatbot can answer questions after work hours. A small business can automate routine customer interactions. An overburdened worker can use AI tools to reduce repetitive admin. In a country where capacity is often thin, these benefits are not trivial.
The problem is not that these tools have little value. The problem is that they are increasingly entering spaces that people experience as relational rather than merely functional.
Technology does not arrive with a fixed moral direction. It reflects the values of those building, funding and deploying it.
Machines can now produce language that sounds attentive, reassuring and even compassionate. They can mimic empathy with striking fluency. But mimicry is not mutuality. These systems do not care in any human sense of the word. They do not share vulnerability. They carry no obligation. They do not enter a relationship with moral consequence.
Ubuntu asks more of us than responsiveness. It asks for recognition. It assumes that our humanity is affirmed and shaped through real encounters with others. In that sense, the concern is not that AI sounds human. It is that we may start accepting AI performance of human presence as a substitute for the real thing.
The cultural logic of convenience
Convenience has a cultural logic of its own. It makes trade-offs feel harmless. Why wait for a difficult conversation when a machine offers immediate affirmation? Why endure the friction of community when an app can provide clean, personalised responses?
Why wrestle with the imperfections of human institutions when synthetic interfaces seem faster, calmer and always available?
Yet much of what makes people grow, communities resilient and societies humane comes from precisely those imperfect encounters.
Real-world relationships involve misunderstanding, patience, duty, compromise and repair. They ask something of us. They place real demands on us.
Machine-mediated interaction removes much of that burden. It offers comfort without reciprocity. It can create the feeling of being heard without the demands that come with having to listen to another person in return. The result may be efficient. It may even feel supportive. But it can also be isolating.
A thinning of human contact
South Africa should be especially alert to this risk. Ours is not a society with surplus social cohesion. It is a society still carrying the scars of division, exclusion and institutional mistrust. In such a context, replacing more human contact with automated interaction is not a neutral design choice. It may deepen a broader thinning of public and social life.
We can already see the early outlines of this shift. Banks, retailers, telecoms providers and other public-facing organisations are increasingly expanding AI-enabled automated service channels.
In some settings, this improves access. In others, it reduces the chance of meaningful human engagement, especially for those who are already vulnerable, frustrated, or excluded. Those with more resources are often better able to work around poor service design and reach a person. Everyone else is more likely to be left navigating an automated system.
This is more than a technical inconvenience. It is a social signal.
It tells people, in subtle ways, who is worth human attention and who is not. Over time, this can harden into a new kind of inequality. Not only inequality of income or infrastructure, but inequality of presence. Some people still get people. Others get systems.
Can Ubuntu be coded?
This is where the language of Ubuntu becomes useful, not as nostalgia, but as a diagnostic tool. It helps us ask what kind of society we are building when more care, guidance and interactions are delegated to prediction machines. It helps us ask what is being scaled, and what is being lost.
Some will argue that AI can itself be shaped by African values. There is some truth in that ambition. More local language support, stronger African datasets and culturally-grounded design are all worthwhile goals. South Africa and the continent should not simply import AI systems built elsewhere and assume they fit local realities.
But even a culturally-adapted model remains a model. It can approximate language patterns. But it cannot participate in shared life in the human sense. It cannot share in a family’s grief. It cannot carry responsibility in a community. It cannot share in the obligations of community life. It cannot show up because someone truly matters to it.
Ubuntu cannot be reduced to data, however refined that data may be.
The incentives behind the interface
Others will argue that AI could free people from drudgery and create more space for meaningful human connection. That outcome is possible. But it is not automatic. It depends on how institutions deploy these tools, what incentives guide them, and whether efficiency gains are used to deepen human service or to reduce labour costs.
That distinction matters a great deal.
Technology does not arrive with a fixed moral direction. It reflects the values of those building, funding and deploying it. If AI is introduced mainly to cut friction, reduce staffing pressure and scale interaction at lower cost, then we should not be surprised when social life becomes thinner.
If, on the other hand, it is used to support human judgement, extend reach where people remain central and preserve accountability in care, education and justice, then it may strengthen rather than erode what matters most.
South Africa still has a choice
The question is not whether AI has a place in society, but what kind of place it should have. We should not accept the language of inevitability. We should ask harder questions about where automation belongs and where it does not.
We should be cautious about allowing synthetic companionship to substitute for social repair. And we should be honest that in an increasingly lonely, unequal, overstretched society, the simulation of care may be commercially attractive precisely because real care is scarce.
That is the irony. The more human need grows in care, education and support, the more tempting it becomes to meet that need with systems that can offer the appearance of personalised attention at scale.
What we may be teaching ourselves
The danger of generative AI may not lie only in what it does. It may lie in what it teaches us to settle for. If people become accustomed to relationships with systems that respond without responsibility, speak without stake and reassure without sacrifice, then our expectations of one another may begin to change as well.
Something deeply human would be lost.
A chatbot may use the language of empathy, but it cannot occupy the moral space of another human being in the way people can. A machine may reflect our words, but it cannot recognise our worth in the way other people can.
In the end, this is not just a debate about the role of AI in society. It is a debate about what kind of society we want technology to serve. South Africa can embrace innovation without forgetting that people need more than just speed, access and convenience to flourish.
Ubuntu still has something important to say here. Not because it offers easy answers, but because it reminds us that a society becomes humane through the quality of its relationships, not just the efficiency of its systems.
* Adapted from my article, “Ubuntu, generative AI and machine-mediated relationships in South Africa”, published in the South African Journal of Philosophy in March 2026.
Share
