BY IDO LEKOTA
These AI-driven adverts are ubiquity on steroids. They flood household television screens with promises of frictionless efficiency, effortless productivity, and lives liberated from inconvenience. They essentially present artificial intelligence as the indispensable assistant: always available, never tired, endlessly analytical, and incapable of the emotional complications that supposedly slow human beings down. Their message is seductive. AI will help us work faster, communicate better, make smarter decisions, and eliminate the small inefficiencies that clutter our professional and personal lives.
Unfortunately, beneath this polished promise lies a troubling proposition: that anything associated with human behaviour – hesitation, conversation, courtesy, uncertainty, empathy, and reflection—is an obstacle to be removed. Even more disconcerting is the cultural domination embedded in such adverts, whereby the corporate style of the countries in which these AI systems are developed is presented as a universal model of efficiency. This is particularly troubling when such technologies are marketed in countries such as South Africa, where foreign-designed systems may be imposed on societies with different histories, cultures, languages, and social priorities.
One such advert offers a seemingly innocuous instruction: “Never – I mean never – start a meeting by saying, ‘Good morning, all.’” The statement is delivered with the confidence of a productivity commandment. A simple greeting, the commercial implies, is wasteful. It consumes time without producing measurable value. It delays the meeting’s “real” purpose. In the world imagined by the advert, the ideal worker should move immediately to the agenda, the target, the transaction, and the outcome.
There is something disturbing about this apparently harmless advice. A greeting is not merely a collection of words exchanged before business begins. It is an acknowledgement that other people are present. It recognises colleagues as human beings rather than functions, departments, data points, or instruments of production. It creates, however briefly, a social space in which people are seen and heard. To remove it in the name of efficiency may save a few seconds, but it also expresses a wider tendency: the reduction of human interaction to its most economically useful components. Artificial intelligence promises greater efficiency, but efficiency without humanity can become another name for exclusion, surveillance, control, and disposability.
The central concern is not that machines are becoming more capable. It is that institutions, employers, and societies may increasingly begin to organise human life according to the logic of machines.
The anxiety is not confined to one advert, one corporation, or one country. It is universal in scope because the systems being developed and deployed by artificial intelligence are rapidly entering almost every sphere of life. They influence how people work, learn, communicate, receive medical attention, access financial services, apply for employment, interact with public institutions, and understand themselves.
In each of these spaces, AI can be used to assist human judgement. But it can also displace it, narrow it, or make it invisible.
Research conducted last year at the universities of Wrexham and Gloucestershire, in the United Kingdom, investigated how the rapid evolution of AI, digital technology, and social media has reshaped cognitive functions – including memory, attention, language, problem-solving, decision-making, and social cognition (i.e., how people store and apply information about other people and situations). This research directly relates to emotional intelligence and how people relate to others in given situations, highlighting both the cognitive benefits and risks of AI-mediated interaction.
The research found that while AI, digital technology, and social media enhance learning, problem-solving, and efficiency, they also pose risks such as cognitive offloading, distraction, and reduced critical thinking. It, for example, revealed a negative correlation between frequent AI use and critical thinking skills, suggesting over-reliance on AI diminishes independent reasoning. Most importantly, the research findings showed “that digital media has reshaped social cognition, altering empathy, social skills, and interpersonal interactions. Excessive screen time reduces real-world engagement, making it harder to read non-verbal cues and process emotions. Adolescents with high digital media consumption also showed lower empathy and weaker social perspective-taking skills.”
These findings are not abstract. They are actually about a cultural shift in which AI-driven adverts and platforms actively trivialise social etiquette in the name of efficiency. When AI systems are marketed as tools that liberate us from the “burden” of greetings and small talk, they do more than change how we communicate; they recode what we value. Efficiency becomes virtue; relationality becomes friction. This is not neutral; it is a form of cultural domination, where the logic of capital – speed, output, transaction – colonises the intimate spaces of human interaction. This is the essence of dehumanisation by AI.
In South Africa, this cultural domination collides with a society still grappling with the legacy of apartheid and the unfinished project of nation-building. The Constitution enshrines human dignity, equality, and non-racialism as foundational values. The philosophy of Ubuntu – “I am because we are” – reminds us that our humanity is bound up in the humanity of others. Yet the seduction of efficiency, marketed as progress, threatens to unravel these hard-won gains. When AI systems encourage us to skip greetings, to treat people as data points, and to prioritise output over relationship, they undermine the social fabric that holds a diverse and unequal society together.
The paradox is stark: in a country where relational apartheid already limits meaningful human engagement for the marginalised, AI-driven dehumanisation risks deepening this exclusion. The poor are not only excluded from wealth and opportunity; they are excluded from the rituals of mutual recognition that signal respect and shared humanity.
When AI systems teach users to treat human interaction as inefficiency, they normalise emotional detachment and encourage users to treat people as data points, not as moral agents. In South Africa, where post-apartheid nation-building has depended on social solidarity and everyday acts of acknowledgment, this is not just a technological issue; it is a moral and political crisis.
Yet, as the stakes rise, the legislative response lags. South Africa’s draft National AI Policy, gazetted for public comment in April 2026, was withdrawn just weeks later after it emerged that at least 10% of its academic references were fictitious and potentially AI-generated. The Minister of Communications and Digital Technologies, Solly Malatsi, has announced a revised policy – now targeted for January 2027, with full implementation not expected before 2027/2028.
In the meantime, AI regulation in South Africa remains fragmented and indirect, governed by general legislation such as POPIA, the Cybercrimes Act, and the Consumer Protection Act, but with no dedicated AI statute
Far from being merely bureaucratic delay, the legislative impasse suggests a failure of political imagination. While the European Union has adopted the AI Act (phasing in through 2027), and jurisdictions such as China and Brazil have implemented targeted AI regulations, South Africa remains in a pre-legislative, policy-development phase. The draft policy envisions a three-year phased implementation roadmap, with Year 1 (2025/26) focused on finalising the policy framework and addressing “unacceptable risks,” Year 2 (2026/27) on developing sector strategies and guidelines, and Year 3 (2027/28) on full implementation of governance and regulatory measures. The withdrawal of the draft, and the scandal over AI-fabricated references, has exposed the weakness of institutional capacity and the risk of regulatory capture by the very technologies that need oversight.
The result is a regulatory vacuum in which AI-driven dehumanisation can flourish unchecked. Without binding rules on algorithmic transparency, data justice, or the ethical design of AI systems, the market logic of efficiency and extraction could continue to dominate. The poor and marginalised could remain disproportionately surveilled, their data extracted and monetised, their agency eroded.
South Africa’s encounter with AI cannot be separated from this history. The country’s present inequalities are rooted in colonialism and apartheid, systems that classified people according to race, controlled their movement, restricted their access to education and employment, and organised economic life around the interests of a minority. The democratic transition dismantled the formal legal architecture of apartheid, but it did not eliminate the structural inequalities embedded in land ownership, wealth, education, geography, and employment.
This history matters because technologies do not arrive in an empty social landscape. They are introduced into institutions that already possess unequal power.
They draw on data produced by societies in which some communities have been over-policed, under-served, misrepresented, or excluded from formal systems altogether. If these historical patterns are reproduced in algorithmic systems, AI may give old injustices a new appearance of scientific legitimacy.
This is particularly important in a country facing high unemployment and widespread youth exclusion. For many South Africans, work is not merely a source of income. It is connected to dignity, identity, social recognition, and the ability to support families and communities. If AI is introduced primarily as a mechanism for reducing labour costs, increasing surveillance, or replacing entry-level opportunities, it may deepen insecurity rather than create inclusive prosperity.
The question, then, is not whether South Africa should reject AI. That would be neither realistic nor desirable. Artificial intelligence could support better healthcare, improve educational access, strengthen agricultural productivity, assist small businesses, expand financial inclusion, and help public institutions respond more effectively to citizens. It could assist workers with disabilities, enable entrepreneurs to reach new markets, and create opportunities for people who have historically been excluded from certain industries.
But these possibilities will not emerge automatically from technological innovation. They depend on the values that guide the development and use of AI. A system designed solely to maximise profit will produce different outcomes from one designed to advance social welfare. A company that treats workers as disposable costs will use AI differently from an organisation that regards employees as partners in innovation. A government that sees citizens primarily as risks to be managed will deploy technology differently from one that views them as rights-bearing participants in democracy. That is the challenge that SA faces.
