Speculative fiction has long served as a mirror for our technological ambitions—sometimes a funhouse mirror, sometimes a sobering pane of glass. The ongoing debate over what separates science fiction from fantasy is less about genre purity and more about our collective hopes and anxieties regarding the future. When it comes to technology, especially AI, understanding this distinction is more than a parlour game; it’s a toolkit for separating plausible innovation from wishful thinking[1][5].
Consider the latest wave of AI hype: the promise of “agentic digital twins” for IT professionals. On the surface, it’s a tempting prospect—imagine delegating all your tedious tasks to a digital doppelgänger. But beneath lies a thicket of unanswered questions: Who’s accountable when the twin makes a mistake? What happens to this digital entity when you change jobs? These aren’t just technical questions, but philosophical and ethical ones, reminiscent of the Sorcerer’s Apprentice—where magical shortcuts inevitably lead to unintended chaos[5].
This isn’t the first time the tech world has chased after digital intelligence. The 1980s saw the rise and fall of expert systems, which aimed to codify human expertise into software. Back then, as now, the technology was promising, the investment was massive, and the results were underwhelming. The core issue wasn’t a lack of computing power—Moore’s Law was in full swing—but the fact that human expertise is messy, intuitive, and deeply personal. Knowledge isn’t just data; it’s experience, context, and sometimes, serendipity. That’s why even today, after years of schooling, graduates often find themselves floundering in their first jobs. AI, for all its advances, still struggles to replicate the nuance and flexibility of human reasoning[5].
Fast forward to 2025: Large language models (LLMs) and large reasoning models (LRMs) are the new frontier. Research, such as that published by Apple, shows that while these models can handle simple and even moderately complex tasks, they falter when the problems become truly intricate. In fact, as problem complexity increases, the reasoning abilities of these models can actually regress—sometimes giving up entirely, even when provided with the correct algorithms. This suggests that there may be hard limits to how far AI can scale as a generalized reasoning machine. The illusion of intelligence is often just that—an illusion, reinforced by clever marketing and our own tendency to anthropomorphize technology[5].
The distinction between science fiction and fantasy helps us navigate these murky waters. Science fiction presents scenarios that, while not real now, could plausibly become so as technology advances—like self-aware digital twins that evolve over decades. Fantasy, by contrast, imagines the impossible—things that cannot happen without a suspension of the laws of nature, like dragon-riding pixies or a Gandalf-led AI team[5][1]. In the context of AI, most current claims fall somewhere between these poles, often veering closer to fantasy than their proponents might admit.
For those working in IT, this isn’t just an academic exercise. Developers are on the front lines, testing the boundaries of what’s possible and reporting back when reality falls short of the hype. Their work is critical not just for the industry, but for society as a whole, as we decide which technological promises to chase and which to leave as the stuff of dreams.
Ultimately, whether an idea is science fiction or fantasy isn’t just about definitions—it’s about responsibility. As we push the limits of AI, we must remain vigilant, skeptical, and honest about what’s truly possible. The future is built not just on code and data, but on the stories we tell about what’s coming next—and which of those stories we choose to believe.
