‘There probably is a critical point in the history of mankind beyond which human affairs, as we know them, could not continue.”
When Sam Altman remarked this week that “we’re now, like, in the singularity,” he revived a word that has travelled an extraordinary intellectual journey. The discussion that followed was less remarkable for its intensity than for the fact that people appeared to be debating entirely different ideas while using exactly the same word. For some, the singularity means superintelligent machines. For others, it describes a technological tipping point. Still others believe it is little more than a science fiction fantasy.
The confusion is understandable because the singularity did not originate in artificial intelligence. It began in astrophysics, a subject that has fascinated me since high school. Long before I studied medicine or entered the pharmaceutical industry, I spent countless hours reading about black holes and Einstein’s description of the universe. Back then, singularity belonged exclusively to cosmology. Seeing the same word dominate discussions about artificial intelligence decades later felt like watching two completely different worlds unexpectedly converge.
In astrophysics, a singularity is not simply the center of a black hole. It represents the point at which our mathematical models cease to provide reliable predictions. According to Einstein’s General Theory of Relativity, the collapse of a sufficiently massive star leads to conditions so extreme that the equations themselves reach the limits of what they can explain. Most physicists do not interpret this as a failure of nature, but as evidence that our understanding has reached its boundary and that a more complete theory must eventually replace it.
That idea eventually escaped physics. John von Neumann, one of the founders of modern computing, was probably the first to borrow the concept as a metaphor for technological progress. Shortly before his death in 1957, his friend Stanislaw Ulam recalled Von Neumann wondering whether the accelerating pace of technological change might approach “some essential singularity in the history of the race beyond which human affairs, as we know them, could not continue.” Decades before personal computers, the internet or artificial intelligence, Von Neumann had already begun asking whether technological acceleration might eventually make the future impossible to predict using the assumptions of the past.
That intuition was later developed by computer scientist and science fiction writer Vernor Vinge. In his influential 1993 essay The Coming Technological Singularity, Vinge argued that once machines became capable of designing increasingly capable machines, technological progress could accelerate beyond humanity’s ability to anticipate its consequences. Ray Kurzweil later introduced the idea to a much broader audience, but the central question remained essentially the same: what happens when intelligence itself becomes the engine driving future intelligence?
Whether that moment has already arrived is precisely what Sam Altman’s recent remark has placed back at the center of the conversation. His statement was not that artificial intelligence has suddenly become conscious or that humans have become obsolete. Rather, it reflected the observation that AI systems are increasingly helping researchers write software, optimize algorithms, analyze scientific literature and accelerate the development of future AI models. Human intelligence is still leading the process, but it is no longer the only force contributing to technological progress.
Whether that deserves to be called a singularity is ultimately a matter of definition. The more interesting question is what follows if the underlying idea is correct. Every description of the technological singularity assumes that intelligence can improve itself. Much less attention has been given to a different requirement: every improvement requires an objective. Evolution improves organisms because survival and reproduction provide a criterion for success. Engineers improve aircraft because efficiency, safety and performance can be measured. Artificial intelligence is no different. It cannot improve itself in the abstract. It can only improve according to the objective it has been asked to optimize.
This is where Stuart Russell offers perhaps the most important contribution to the debate. Russell, professor of Computer Science at the University of California, Berkeley, and co-author of the field’s most widely used textbook, argues that intelligence is not an objective in itself. Intelligence is simply the ability to achieve an objective efficiently. As systems become increasingly capable, the defining question gradually shifts from How intelligent can they become? to Who decides what they should optimize?
Nick Bostrom illustrated the same problem through his now famous paperclip thought experiment. An extremely intelligent machine instructed to maximize paperclip production would not threaten humanity because it had become evil. It would threaten humanity because it would pursue its objective exactly as instructed, revealing that humans had failed to define the objective as carefully as they imagined. Intelligence amplifies objectives; it does not judge whether those objectives are wise.
Seen from that perspective, the singularity becomes something more profound than a prediction about artificial intelligence. In astrophysics, a singularity marks the point where our theories stop explaining reality. In artificial intelligence, it may eventually describe the moment when human intelligence is no longer the only engine driving technological progress. If that transition is beginning, then intelligence itself may become increasingly abundant. The scarce resource will not be computing power or algorithms, but human judgment: the ability to define the objectives that increasingly intelligent systems should pursue and the constraints they should never be allowed to ignore.
The future of science may depend less on our ability to produce answers than on our wisdom in choosing the questions that deserve them. And in healthcare, those choices are never merely technical. Deciding which diseases deserve priority, which risks are acceptable, which technologies should be developed first and which scientific frontiers are worth crossing is, ultimately, deciding the future we want to build.
For the first time in history, humanity may find itself in a position where intelligence is no longer the primary constraint on scientific progress. If that happens, our greatest responsibility will not be creating more intelligence, but exercising better judgment. The decisions we make about which problems deserve to be solved may shape not only the future of medicine, but the future of our species.
Perhaps that is the deepest implication of the singularity. It is not that machines become more intelligent than humans. It is that humanity may gradually begin to think less as billions of individuals solving isolated problems and more as a civilization consciously choosing the direction of its own future.
This article was originally written in Spanish and translated into English with the assistance of artificial intelligence.
