The 20th-Century Crisis in Mathematical Foundations: From Axioms to Gödel and Type Theory

The twentieth century shook the foundations of mathematics to its core — but where did it all lead?

It began with a desire to build a rigorous foundation for the most rigorous of all sciences. By the nineteenth century, it had become entirely clear that physics, chemistry, and biology rely on experiment as their primary means of verification. But mathematics is different. No empirical observation can prove that there are infinitely many prime numbers, and a brute-force proof of Fermat’s Last Theorem is fundamentally impossible.

Mathematics operates on strict rules — axioms. But the central question remained: which axioms, exactly? And who decided they were admissible in the first place?

The First Cracks in the Foundations of Mathematics

Attempts to discard Euclid’s parallel postulate led to entirely new, internally consistent geometries — those of Lobachevsky, Riemann, and others. This was already a warning sign: rigorous mathematics could exist in multiple versions.

It soon became clear that similar problems lurked at the very foundation of mathematics itself — in arithmetic and logic. Paradoxes emerged, such as the barber paradox, which seemed to undermine the very idea of rigorous reasoning.

Consider this for a moment: “This statement is false.” Is it true or false?

Mathematics needed a definitive answer. Four major schools of thought took shape.

Four Schools of Mathematical Foundations

Formalism, led by David Hilbert, held that any mathematical statement could in principle be proven, provided one chose the right axioms and inference rules.

Logicism, advanced by Bertrand Russell and Alfred North Whitehead, sought to reduce all of mathematics to logic — resolving paradoxes at the cost of artificial restrictions and elaborate constructions.

Intuitionism, founded by Brouwer, argued that only what can be explicitly constructed exists; the law of excluded middle is not universal, and truth cannot be separated from construction.

The set-theoretic approach, initiated by Cantor and developed by Zermelo and Fraenkel into what became ZFC, centered on a hierarchy of sets, the axiom of choice, and the continuum hypothesis.

Gödel’s Decisive Blow to Axiomatic Systems

By the 1930s, Kurt Gödel had delivered a decisive strike. Any sufficiently powerful axiomatic system is incomplete: there will always exist statements that can neither be proven nor refuted within that system.

Löwenheim and Skolem demonstrated that a single axiomatic system can admit fundamentally different interpretations. And Paul Cohen, by proving the independence of the axiom of choice and the continuum hypothesis from the remaining Zermelo–Fraenkel axioms, effectively showed that there is not one mathematics, but many.

It turned out that mathematics rests on no firmer ground than physics or chemistry. Where absolute rigor had been expected, what was found instead was a sprawling landscape of disconnected axiomatic systems — each useful, each limited, and each harboring its own unprovable statements.

What the Last Hundred Years Have Produced

Today, ZFC remains the official standard — simply because it is convenient and familiar. But competitors have grown up around it. One of the most promising is type theory. Instead of sets, it works with types, where statements and constructions are inseparable. A proof here is simultaneously an object, and logic is tightly bound to computation. In homotopy type theory, equality is understood as structure rather than formal coincidence, which makes it particularly well-suited for computer-verified proofs.

Mathematical Foundations in the Age of AI

As artificial intelligence and formal systems become ever more central to modern work, mathematics is naturally gravitating back toward constructive and intuitionist approaches. There is little time for lengthy philosophical debate about foundations when a result from a large language model needs to be verified right now. And that is precisely why the present — and the future — looks especially interesting.

Even so, it is hard not to side with Hilbert in the end: in mathematics, there are no limits to knowledge beyond the limits of the researcher’s own talent.