Riemann Hypothesis

Prime numbers are numbers that are only divisible by themselves and by 1: 2, 3, 5, 7, 11, 13, and so on. They appear quite irregularly among the numbers. Sometimes two lie close together, sometimes there is a large gap. The German mathematician Bernhard Riemann formulated a conjecture in 1859 about how regular this distribution actually is at its core. His statement concerns a computational rule that he studied, and the points at which this rule yields the value zero. Riemann claimed that all the interesting ones of these zeros lie exactly on a single line. To this day, no one has proven this and no one has found a counterexample.

A Million Dollars for a Proof

The Riemann Hypothesis is on the list of the seven Millennium Problems. This list was published by the Clay Mathematics Institute in the year 2000. For each solved problem there is a prize of one million US dollars. So far, only one of them has been cracked, and it was not this one.

However, the money is not the real reason for its fame. Many hundreds of mathematical theorems have already been proven under the assumption that the hypothesis is true. Such results are called conditional. If the hypothesis were ever disproven, a large part of this structure would collapse. Conversely, a proof would secure all these theorems at once.

A common misconception is that a proof would immediately break encryption on the internet. That is false. Methods like RSA rely on the fact that factoring large numbers into prime factors takes an extremely long time. The Riemann Hypothesis says something about the distribution of prime numbers, not about how to find them quickly. It provides no recipe for cracking codes.

Zeros on the Critical Line

At the center is the Riemann zeta function. A function is simply a computational rule: you put in a number and get a number out. This particular function arises by adding infinitely many fractional parts. Riemann extended it to so-called complex numbers. These are numbers with two components that can be represented like points on a plane.

What is interesting are the zeros, that is, the inputs for which the result is exactly zero. Some of these are trivial and easy to determine. The remaining ones all lie within a vertical strip of this number plane. Riemann’s conjecture states that they even crowd together on a single line right in the middle of this strip. This line is called the critical line.

The connection to prime numbers is astonishingly close. One can imagine the zeros as the fundamental tones of an instrument. Combined, they produce the irregular pattern of the prime numbers. If all the tones lie neatly on a line, then the number of primes fluctuates only slightly around a smooth average value. If a zero were off the line, there would be an unexpectedly large irregularity. Computers have already checked more than ten trillion zeros, all of which lay on the line. This is not a proof, however, since there are infinitely many.

Why AI Researchers Talk About It

In tech news, the Riemann Hypothesis has appeared as a benchmark for several years now. When companies present their new AI systems, they like to cite unsolved mathematical problems as a long-term goal. A language model that independently finds such a proof is considered evidence of genuine mathematical reasoning. So far, this has not happened. Critics view the example as marketing, since even experts have been failing at this problem for over 160 years.

A more realistic role is that of software as a tool. Proof assistants like Lean check mathematical arguments step by step for errors. Large computer programs calculate zeros with high precision. AI systems are now being trained to suggest proof steps that a human then verifies. For smaller theorems, this already works.

Outside of research, one encounters the term in films, novels, and puzzle competitions. Alleged proofs regularly circulate on the internet. Almost all of them come from amateurs and contain errors. Even renowned mathematicians have tried and failed. This is precisely what makes the hypothesis a fixed point of reference whenever it is claimed somewhere that a machine now thinks like a human.