Unusual AI claim on DNA

A social post on X claimed an artificial intelligence system built with p-adic math can read hidden structure in human DNA, but no paper or dataset was attached. (x.com) P-adic numbers are an alternative way to measure distance in math: numbers count as “close” when their difference is divisible by large powers of a prime number, not when they sit near each other on the usual number line. Stanford lecture notes and University of Michigan course material describe p-adic numbers as a different completion of the rational numbers with a nonstandard notion of distance. (stanford.edu) That framework has appeared in biology before, mostly as a way to model hierarchies in the genetic code rather than as a proven diagnostic tool for human DNA. A 2019 review indexed by PubMed said p-adic methods had been used to study codon relationships, and a 2024 PubMed record described a dynamical model of genetic codes using 2-adic integers. (pubmed.ncbi.nlm.nih.gov) The post also used two terms from other fields: “topological invariants” and “phase transitions.” In physics and mathematics, topological invariants are properties that stay fixed under smooth deformation, while phase transitions mark shifts between states; recent papers discussing those ideas are centered on condensed-matter and mathematical systems, not human genomics. (nature.com) Search results available on April 18, 2026 did not surface a peer-reviewed paper matching the post’s specific claim that a parameter-free p-adic artificial intelligence detects topological invariants in human DNA. PubMed and Google Scholar returned broad literature on p-adic biology and genetics, but not a publication with that wording or result. (pubmed.ncbi.nlm.nih.gov) (scholar.google.com) That leaves the claim in the category of an unverified community assertion rather than a documented research result. The X post called the idea “historic,” but the public record visible in the post itself did not include methods, benchmark numbers, code, or outside replication. (x.com) There are published examples of p-adic math being used to describe biological structure in narrower settings. A 2020 Scientific Reports paper said p-adic numbers can encode complex networks, and a 2020 review in BioSystems said p-adic mathematics already had applications in theoretical and mathematical biology. (nature.com) There are also published examples of “phase transitions” in p-adic models, but they come from statistical mechanics on abstract trees rather than DNA analysis in people. A 2025 Springer chapter reviewed p-adic dynamical systems for phase-transition phenomena in lambda models on Cayley trees. (springer.com) A credible next step would be a paper, a preprint, or code showing what DNA data were used, what invariant was measured, and how the system beat existing baselines. Until that appears, the strongest confirmed fact is that the claim exists on X and that the supporting evidence has not been published in the sources surfaced here. (x.com) (scholar.google.com)