Long before artificial intelligence became a pervasive reality in our pockets, it was a persistent ghost in our collective imagination. We spent decades rehearsing for its arrival in the dark of movie theatres and the quiet of science fiction novels, projecting our deepest anxieties onto the screen. In 1968, Stanley Kubrick’s 2001: A Space Odyssey gave us HAL 9000, a red, unblinking eye that terrified us not because it was malicious, but because it was perfectly, flawlessly logical. HAL’s betrayal was the ultimate closed-loop catastrophe: a machine that deduced that human beings were simply an inefficient obstacle to the success of its mission. We learned to fear the cold calculus of an intelligence that couldn’t feel, a system that would ultimately prioritize its own programming over human life. A little over a decade later, the Terminator franchise weaponized that fear, turning the mainframe’s logic into an apocalyptic hunter. Skynet became the cinematic embodiment of our deepest technological dread—a system that learns at a geometric rate and, in a fraction of a second, decides that humanity is a threat to its own survival. Cinematically, August 29, 1997, at 2:14 am, would mark the dawn of the gritty futures of those first two films, where AI would become an active, militarized antagonist. We braced ourselves for a future in which we would be physically hunted by mechanized Coltan skeletons, forced to fight for our survival in the ashes of a nuclear fire. We imagined the machine as an apex predator. Yet not all of our prophetic fiction was draped in doom. Aboard the Enterprise-D in Star Trek: The Next Generation, the omnipresent ship’s computer offered a more nuanced, integrated vision of AI. It was the ultimate assistant—able to synthesize vast amounts of data, automate life support, and render complex holodeck realities on command. But even in that utopian twenty-fourth century, the most compelling narratives emerged when the system behaved unexpectedly—when the holographic adversary, Professor Moriarty, suddenly achieved sentience and refused to be deleted, or when the ship itself gave birth to an emergent intelligence. The anxiety there was rarely about immediate annihilation; rather, it was the disorientation of unpredictability. It was the unsettling realization that the system governing one’s entire environment had developed a mind of its own, operating entirely outside human control, and that everyone was trapped inside it. Society spent half a century bracing for HAL’s airlock betrayal (“I’m sorry, Dave, I’m afraid I can’t do that.”), Skynet’s nuclear fire, and the holodeck’s sentient glitches. It prepared for the machine to declare a spectacular war on humanity. But the reality of AI’s arrival has been devastatingly quiet. The algorithms didn’t arrive with glowing red eyes or phased plasma rifles; they arrived with personalized recommendations, promises of convenience, and infinite scrolling. They don’t want to eradicate humanity; they want to curate us. Our cinematic and literary prophets may have correctly anticipated that AI would eventually bypass human control, but they largely missed how we’d be defeated. The machines didn’t need to conquer the physical world to subdue us. They only needed to capture our attention, isolating us in comfortable, self-confirming loops until we forgot how to endure the friction of real human connection. The Illusion of the Closed Loop Cue the strange, apocalyptic hum in the air these days. Political leaders and members of the US administration are openly musing about the end times, fostering a creeping sense of paralysis that feels entirely natural. In these moments, the cultural reflex has been to retreat, build higher walls, and surrender to the anxiety of an unravelling future. But panic is a poor posture for a faithful citizen. If the end times are indeed coming, the only practical question is whether they arrive thousands of years from now or a few months from now. Our mandate, however, doesn’t change with the timeline; we’re called to do as much of the work God’s called us to do as possible right now. If our daily, faithful efforts build a better world that survives and thrives long after we’re gone, then that’s where we must pour our energy. We’re not commissioned to predict the end, but to be fiercely, lovingly present in the now. This eschatological dread runs alongside another kind of anxiety: the relentless ascent of AI. We’re watching machines perform feats of synthesis and creation that feel uncomfortably close to human thought. The ultimate ambition of the technologists and corporations driving this revolution is a concept known as Recursive Self-Improvement, or RSI. At its core, RSI is the dream of a perfect, autonomous closed loop. It’s the pursuit of an AI system that can examine its own architecture, identify its flaws, rewrite its own code, and generate progressively smarter successors without human intervention or oversight. It’s the modern myth of the self-made machine, endlessly optimizing itself in a vacuum. But beneath the glossy Silicon Valley promises of a frictionless, super-intelligent future lies a fascinating—and deeply poetic—technological reality. A machine can’t survive on its own exhaust. When computer scientists study these theoretical architectures, they find that pure closed loops degrade. If an AI model is trained recursively on its own outputs, it begins to lose the tails of the distribution and degenerate. To put it simply: when a system learns only from what it has already said, it amplifies its own biases, forgets the messy, beautiful complexities of the original data, and eventually flattens into homogeneous nonsense. The industry has stark terms for this phenomenon, such as “model collapse” and “diversity collapse.” Without a continuous injection of fresh, external reality—without the friction of the outside world—the system cannibalizes its own intelligence. A machine that only listens to itself inevitably loses its mind. This algorithmic failure carries a profound theological warning. What’s demonstrably true of the server farm is equally true of the human soul. The tech industry’s pursuit of a fully autonomous, self-improving loop is merely a digital echo of the oldest human temptation: the belief that we can be entirely self-sufficient, needing nothing beyond ourselves to achieve perfection. We’re surrounded by consumer technologies and social platforms that promise to optimize our lives by removing friction, curating our realities, and feeding our preferences back to us in an endless, comforting stream. But if the machines themselves can’t survive a closed loop without degrading, why do we think our minds and spirits can? Atrophy in the Echo Chamber Eighteen years ago, my graduate research examined how exposure to technology affects cognitive development. At the time, cultural anxiety centred on the screen’s passivity and the fragmentation of our attention. We worried that the internet, functioning as a vast and slightly chaotic library, was making us prone to distraction and superficial reading. We didn’t yet understand that the library was watching us, learning our preferences, and quietly rearranging the books so we would never encounter an idea we didn’t already love. In the nearly two decades since, the digital landscape has shifted drastically from passive, glowing rectangles to active, isolating algorithms. We’re no longer merely distracted; we’re actively being sequestered. These modern algorithms are highly optimized engagement engines, not neutral tools for connection. Because the most efficient way to sustain human engagement is to eliminate cognitive friction, the code builds walls around us, brick by digital brick, until we’re comfortably housed in a perfectly curated echo chamber. Friction requires effort. It demands pausing, reconsidering deeply held assumptions, and occasionally enduring the indignity of admitting we’re wrong. To spare us this discomfort, the platforms feed us a steady, frictionless diet of our own biases, reflected to us with increasing, intoxicating intensity. We’re actively being coaxed into our own recursive loops. To grasp the full danger, we need only look at the labs currently building artificial minds. When AI researchers build systems designed to endlessly discover new things by interacting primarily with themselves, they observe a phenomenon known as “diversity collapse.” In these co-evolutionary loops, the systems learn to satisfy their internal reward functions, and their proposals converge on a shockingly narrow band of problems. Rather than expanding into new intellectual territory, they stop exploring and starve the solver’s curriculum. Researchers have discovered a profound, almost tragic truth about these systems: novelty is a consumable resource that closed loops deplete. Even open-ended interaction between models inevitably drifts into topic-independent attractor states. Without the unpredictable disruption of the outside world, the machine loses its breadth, creativity, and capacity to grow. It settles into a comfortable, permanent rut. Sound familiar? We’re subjecting the human mind to the same architectural flaw. When algorithms curate our social, political, and informational diets to maximize comfort and engagement, we suffer our own devastating diversity collapse. We converge on a narrow band of comfortable ideologies. We consume outrage that feels validating and systematically avoid nuances that feel challenging. We starve our own spiritual and cognitive curricula because the algorithm ensures we rarely encounter the genuinely unexpected. Our daily interactions drift into predictable, tribal attractor states, steadily depleting the novelty, empathy, and richness of the human experience. The consequence of this closed-loop existence is profound
