The Shape of Artificial Intelligence What AI really looks like I. Spooky shapes at a distance The shape of things only becomes legible at a distance. History, for instance, requires temporal distance. The statement “the Western Roman Empire fell in 476 AD” was not a fact in the year 476. It only emerged as a meaningful narrative after centuries of reflection, when scholars could step back and see the slow erosion of institutions, the decline of central authority, and the fragmentation of territory. The deposition of Romulus Augustus, the last Western emperor, was a real event, but its significance as a symbolic end was not immediate. It was only when historians began to zoom in and out across time—compressing decades of political, military, and economic decay into a single, clean endpoint—that the shape of a fallen empire became visible. Distance can also be spatial, not just temporal. In the arid highlands of southern Peru, the Nazca people created massive geoglyphs etched into the desert floor—vast lines, figures, and spirals stretching over miles. From the ground, they appear as little more than random grooves. But from the air, the full forms emerge: a hummingbird, a monkey, a spider, a condor. Only from above can the meaning of the lines be understood, as the human eye, unaided, cannot grasp the totality of the design. AI, too, is still in the process of taking shape. Though the field is approaching its 70th anniversary, the current era of modern AI is barely a decade old. The launch of ChatGPT in 2022, the publication of the transformer architecture in 2017, and AlexNet’s breakthrough in 2012—each a landmark moment—mark the beginning of a rapid evolution. Yet, as with the Nazca Lines, the true form of AI is not yet clear from ground level. We are still too close to see the full picture. The patterns of progress, the shifts in capability, the growing influence of large models on language, vision, and decision-making—these are not yet legible in their totality. The real shape of AI is not a single object, but a sprawling, evolving network of systems, data, and human choices. It is not a machine, but a process. It is not a product, but a practice. And like the geoglyphs, its meaning will only become clear when we step back—when we can see the whole, not just the parts. Until then, we are left with shapes that seem strange, even spooky, from a distance. But that, perhaps, is the first sign of something real.