By Veon Bock
Media & Epistemology
Artificial Intelligence (AI) is a concept that most people are still grappling to understand. I must admit, I find myself increasingly amused by the hullabaloo surrounding generative pre-trained transformers (GPTs), underpinned by large language models. These tools are undoubtedly powerful vectors of change, but I believe we are being unnecessarily alarmist by creating pitfalls where none exist. Someone recently likened their arrival to the invention of the Gutenberg Press, pointing to the advancement of knowledge and social upheaval that followed that seismic shift. Yet such analogies, while provocative, often obscure more than they reveal. In my view, much of the current confusion stems from a fundamental mischaracterisation: these tools are frequently described as either being or possessing artificial intelligence. Nothing could be further from the truth. We are, quite simply, putting the proverbial cart before the horse.
The first issue I wish to explore is that, in truth, these tools should more appropriately be understood as simulacra rather than AI. Let me explain why this distinction matters, by way of an analogy.
In recent history, many of us have used software like Microsoft Excel to construct analytic models, some of them rather sophisticated. Excel has long been a workhorse for professionals engaged in quantitative analysis, be it writing macros, running regressions or building valuation models. Yet, no one ever argued that Excel itself was “intelligent.” We always understood that the integrity of any analysis conducted with Excel was entirely dependent on the quality of our inputs and assumptions. Garbage in, garbage out.
Moreover, Excel’s power lies in its “static” logic. Macros, for example, are built on pre-defined, hard-coded scripts. They do not adapt or reinterpret inputs; they execute fixed sequences of commands. By contrast, GPTs are “dynamic”: they generate responses probabilistically, based on linguistic patterns in their training data and the specific context of each prompt. While a macro in Excel executes logic in numerical form, a GPT simulates language. It does not follow a pre-written script; it constructs one in real time, tailored to the prompt.
This distinction is crucial. GPTs, because of their fluency and generative capacity, create the illusion of thought. And that is precisely where the misunderstanding lies. Their linguistic polish gives the false impression of cognition, coherence, even sentience. It is the GPT’s act of “construction”, the on-the-fly generation of text, that is so often mistaken for intelligence. In truth, they are nothing more than simulacra: what Baudrillard described as “representations that mask the absence of reality” or in this case, the absence of actual reasoning or comprehension. They echo patterns, not thoughts.
A second, but no less important, dimension of our misunderstanding of GPTs echoes the warnings issued decades ago by cultural historian Theodore Roszak. In his landmark work “The Cult of Information”, Roszak cautioned against the uncritical embrace of computational tools and the rising mythology of information. He argued that modern society had begun to mistake the accumulation of data for knowledge, and the manipulation of symbols (the ability to process and output information) for understanding. This conflation, he warned, leads not to enlightenment but to alienation. We become estranged from context, depth, and meaning, adrift in a world where information is abundant but insight is scarce. GPTs, in many ways, are the very embodiment of this condition: vast informational resources arranged in syntactically correct forms, yet devoid of meaning and lacking the interpretive soul that gives language its truth.
What has changed is not the nature of the tools, but the illusion they create. it is not artificial intelligence at all; it is an apparatus of simulation. It is a linguistic hall of mirrors that mimics cognition without possessing it. Some contend that we are entering a new “era of knowledge.” This is misleading, as it commits a category error: it conflates the accumulation and ordering of data, which is devoid of interpretation, understanding, and meaning, with knowledge itself. What we are entering is an era of hyper-production of the semblance of knowledge. This is text that wears the costume of coherence, fluency, and expertise while remaining fundamentally void of referential anchoring. What GPT produces are not contributions to a shared stockpile of insight, but hyperreal artifacts: content that circulates as if it were meaningful, but whose meaning is performative rather than propositional. GPT has no fidelity to reality.
This distinction is now more clearly marked by one of the more sordid consequences of this illusion: it is that individuals who neither read widely nor write meaningfully now produce GPT- generated content, passing it off as intellectual work. In doing so, they not only perpetuate a culture of “scarce insight”, but also raise serious ethical concerns. An example of this may be, the student who submits a GPT-written essay, yet fails a “viva” when asked to defend a key claim. In addition, this misuse of GPT leads to the disappearance of subjectivity itself as it simulates intention, argument and creativity. This illusion of authorship is devoid of the existential weight of choosing, reflecting or risking meaning. To those of us who value rigour, to those who read, who write, who think, such hollowness sticks out like a sore thumb. The “garbage in, garbage out” effect is as real with GPTs as it ever was with Excel. Just as a seasoned finance professional can spot a phoney valuation model a mile away, so too can the critical thinker detect the counterfeit write-up.
Lastly, the real debate is not about whether GPTs are intelligent. They are not. The real issue is access. As I proposed in my “Theory of Financial Inclusion and Social Innovation”, developed during my Master’s research, we must ensure that these technologies align with the five principles outlined in that framework: they must be accessible, affordable, usable, sustainable, and ethical. Without intentional democratisation, we risk allowing this next wave of technological advancement to deepen existing inequalities rather than help resolve them.
In short, GPTs are not the problem. Misunderstanding them is. Let’s not conflate linguistic fluency with intelligence, nor simulation with insight. Let us heed Roszak’s warning: information without wisdom is not progress, it is alienation masquerading as innovation. The tools are neutral. It is how we use them, and who gets to use them well, that will determine their legacy.
Bio: Veon Bock, MPhil (UCT) – is currently pursuing a PhD in Philosophy. He is an independent consultant and is currently in the process of authoring two books, one dealing with the financial exclusion problem in South Africa and the other a novella, which memorialises Prof Adam Small and Dr Neville Alexander.
I am deeply saddened to hear of Veon Bock’s passing on March 26, 2026.
I knew Veon as an intellectual and spiritual soul, a man whose inner life carried both depth and discipline. He loved poetry and jazz, and he approached writing with a rare appreciation for structure and clarity. There was always a quiet integrity in the way he thought, spoke, and expressed himself.
I am publishing this posthumously to honour and celebrate all that he was, and all that he gave through his presence and his mind.
May he continue to shine in the great beyond.
Gillian Schutte. Editor-in-Chief.
