For centuries, pure mathematics has stood as one of the most rarefied, protected sanctuaries of human intellect—an abstract realm where progress is measured not by market utility, but by the quiet cultivation of deep, elegant, and often paradigm-shifting insights. Yet, the serene halls of this ancient discipline have recently been shaken by the rapid intrusion of artificial intelligence, sparking a profound philosophical and practical existential crisis among its practitioners. The catalyst for this localized panic arrived in late May, when the tech giant OpenAI proudly announced that one of its proprietary models had successfully disproved an eighty-year-old combinatorial geometry conjecture originally posed by the legendary Hungarian mathematician Paul Erdős. To some observers, like University of Toronto number theorist Jacob Tsimerman, the achievement was nothing short of brilliant, prompting him to declare that the resulting paper was so impressive he would accept it for publication in any reputable journal without a single hesitation. In the broader public square, this feat was immediately packaged as the dawn of a new era: a flashy, triumphant narrative showing that machines could now outthink the world’s most brilliant human minds in their own backyard. However, behind the celebratory press releases, a growing faction of researchers felt a creeping sense of unease. Harvard mathematician Melanie Matchett Wood quickly pointed out that the AI’s paper inexplicably failed to reference a long, rich history of closely related human ideas already present in the existing literature, highlighting a glaring disconnect between the model’s brute-force computational success and actual, contextualized human understanding. This friction exposed a raw nerve in the academic community, raising a vital question that now haunts the entire discipline: what happens to the soul of mathematics when its most complex problems are solved by black-box algorithms that offer solutions without providing the human comprehension necessary to appreciate them?
In response to this encroaching systemic transformation, a courageous coalition of sixteen prominent mathematicians, working in close consultation with global mathematical bodies, took a defensive stand by publishing the Leiden Declaration on Artificial Intelligence and Mathematics. Led by visionary thinkers such as Dame Ursula Martin, a distinguished mathematician and computer scientist at Oxford, and Rodrigo Ochigame, a meticulous historian of computing at Leiden University, the manifesto serves as a collective provocation meant to wrestle back control of the narrative from the marketing departments of Silicon Valley. The emergence of the Leiden Declaration is a direct reaction to the realization that the AI industry is actively using mathematical triumphs as weaponized public relations, distorting the public’s understanding of what mathematics truly is. When OpenAI threw immense computational power and human engineering resources at a highly specific, eighty-year-old Erdős conjecture, they achieved a remarkable result, but they kept the failure rates, the exact prompts used, the underlying training methodologies, and the staggering computational costs completely hidden behind a corporate veil. Dame Ursula Martin and her co-signatories argue that reducing the rich tapestry of mathematics to a series of highly gamified, self-contained exam-style puzzles completely misunderstands the spirit of the discipline. Mathematics is not merely a production line for solving isolated problems; rather, it is a slow, deeply human cultivation of intuition, aesthetic judgment, shared conceptual vocabulary, and profound ontological wisdom. By reducing the field to a series of optimization hurdles, AI corporations risk creating a superficial ecosystem where the flashy exploitation of localized problems overshadows the long-term, patient pursuit of deep conceptual frameworks that have historically underpinned all of modern scientific progress.
At the very core of this scholarly rebellion is a fundamental clash of civilizations between the egalitarian, idealistic culture of academia and the predatory, extraction-based logic of late-stage digital capitalism. For generations, pure mathematics has operated as a rare, highly cherished “gift economy”—an island of intellectual idealism where researchers freely share their ideas, collaborate across borders without expectation of financial gain, and build upon one another’s work with mutual respect and meticulous citation. Columbia University mathematician Michael Harris, who reads hundreds of graduate applications yearly as director of graduate studies, notes that this preservation of pure, unadulterated idealism is what continues to draw young, brilliant minds to the discipline. However, this open-source paradise is now being systematically strip-mined by massive tech conglomerates like Google DeepMind, OpenAI, Anthropic, and a host of well-funded startups such as Harmonic, Math, Inc., and Axiom Math. These corporations are not investing in mathematics out of a altruistic desire to advance the boundaries of human knowledge; instead, they are using the public intellectual labor of mathematicians as free training fodder to enhance the general reasoning capabilities of their models, which they then package, monetize, and inevitably pitch to venture capitalists and military contractors. This dynamic places contemporary mathematicians in a deeply compromising ethical position. Their lifelong, publicly funded contributions to human thought are being gathered without their consent to train proprietary, closed-source models that may eventually be deployed in lethal autonomous weapons systems or surveillance infrastructures—realities that are completely antithetical to the peaceful, creative spirit of mathematical inquiry.
Furthermore, this corporate colonization of mathematics threatens to introduce devastating systemic inequities and epistemic instability into the scientific community. As journal editors already complain of being completely overwhelmed by a deluge of plausible-sounding, AI-generated papers that turn out to contain subtle, hallucinatory errors that are incredibly difficult and time-consuming for human peer-reviewers to detect, the very integrity of the academic record is being compromised. This deluge of synthetic output threatens to paralyze the traditional mechanisms of scientific validation, which rely entirely on scarce human labor and rigorous, trust-based evaluation. Simultaneously, as Ochigame and his colleagues warn, the immense financial and computational asymmetry between technology companies and public universities threatens to create a deeply unfair caste system within mathematical research. When AI development is prioritized, resources and status will inevitably flow toward research questions that are easily amenable to automated, reinforcement-learning models, while deeper, highly abstract, and non-computable inquiries will find themselves marginalized, underfunded, and intellectually neglected. Researchers who lack access to the planetary-scale server farms owned by a handful of American and Chinese tech monopolies, or those who simply choose to preserve human-centered, pen-and-paper methodology, will find themselves structurally disadvantaged. This shift threatens to undermine the diversity of mathematical thought, replacing a historically decentralized, global community of diverse thinkers with a highly centralized, corporate-controlled pipeline designed to optimize models for impending Initial Public Offerings rather than genuine, enduring human understanding.
To truly understand what is at stake under this technocratic onslaught, one must examine the defining essence of mathematics as a uniquely human cultural expression. The tech industry’s prevailing paradigm reduces mathematical progress to a series of binary victories, as though the field were nothing more than a giant game of chess to be conquered by raw, brute-force search algorithms. But to mathematicians, this view is a profound and dangerous impoverishment of their life’s work. The actual beauty of a mathematical proof does not lie in its mere correctness, but in its explanatory power—the way it reveals hidden connections between seemingly disparate realms of reality, the simplicity of its insights, and the profound aesthetic pleasure it brings to those who finally grasp its inner logic. A computer program can output a thousand-page string of logical deductions that verifies a conjecture, but if no human being can comprehend, synthesize, or build upon those steps, the execution remains a sterile, inert artifact rather than a living piece of mathematics. By outsourcing the act of discovery to proprietary, closed-source neural networks, humans risk losing the very cognitive pathways and intuitive leaps that make the discipline so vibrant. The Leiden Declaration is, at its heart, a plea to preserve this human dimension: the slow, collaborative, and often painful process of wrestling with the unknown, exploring the outer boundaries of human imagination, and sharing that hard-won clarity with a community of peers. If we allow corporations to replace the slow cultivation of human genius with the rapid, automated generation of alienated, unexplainable results, we risk permanently damaging the delicate social and educational institutions that have allowed mathematical curiosity to flourish for millennia.
Ultimately, the mathematicians behind the Leiden Declaration are not advocating for a Luddite retreat from technological progress, nor are they trembling in fear of impending obsolescence. They understand that artificial intelligence, when developed and utilized responsibly, can serve as an incredibly powerful assistant capable of accelerating routine calculations, suggesting novel pathways, and liberating human researchers from mundane cognitive labor. Instead, their manifesto is an urgent call for ethical coexistence, demanding that tech companies abandon their secretive, high-intensity public relations campaigns and commit to the foundational values of the scientific enterprise: transparency, open verification, honest methodology, and mutual respect. Rodrigo Ochigame eloquently reassures us that mathematics, as an ancient and deeply rooted form of human culture, is far too resilient to ever be rendered truly obsolete by silicon chips. The desire to understand the universe through the lens of abstract patterns is a fundamental part of our species’ identity, an endeavor that cannot be automated away any more than we can automate the writing of poetry or the experience of grief. The true danger is not that machines will become too smart, but that we, as a society, will allow a small cartel of profit-driven corporations to diminish our definition of intelligence, convince us to abandon our intellectual sovereignty, and dismantle the communal structures of open science. By standing firm against the relentless hype wave, the authors of the Leiden Declaration are reminding us that the ultimate goal of mathematics is not the sheer accumulation of answers, but the expansion of human consciousness, and that is a journey we must continue to walk on our own two feet.
