{
  "id": 15350212,
  "title": "15-year-old Korean posts AI-assisted proof that a 62-faced shape can't pass through its own copy",
  "url": "https://urgent.news/2026/10/18/15-year-old-korean-posts-ai-assisted-proof-that-a-62-faced-shape-cant",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-18T04:00:00.000Z",
  "source": {
    "name": "VnExpress International",
    "slug": "vnexpress-international",
    "url": "https://e.vnexpress.net/news/news/education/15-year-old-korean-posts-ai-assisted-proof-that-a-62-faced-shape-can-t-pass-through-its-own-copy-5132331.html"
  },
  "original_language": "en",
  "account": "Fifteen-year-old Jung Gi-hyo, a middle school student from Gimpo, Gyeonggi Province, recently shared a paper titled \"The rhombicosidodecahedron is not Rupert\" on research platforms Zenodo and GitHub on October 5. In an interview with the JoongAng Ilbo, Jung explained that he generated code for a complex mathematical problem using Claude and then verified the results. He claimed to have chosen the research topic and direction on his own, spending around 38 hours calculating on his laptop.\n\nThe rhombicosidodecahedron, an Archimedean solid with 20 equilateral triangles, 30 squares, and 12 regular pentagons, is believed to possess the Rupert property, allowing a tunnel wide enough for an identical copy to pass through. Mathematicians have widely accepted this property for every convex polyhedron. However, in August 2025, Jakob Steininger and Sergey Yurkevich disproved this by constructing a 90-vertex solid, dubbed the Noperthedron, which lacks such a passage. Jung's paper addresses a narrower conjecture raised by Steininger and Yurkevich in 2021.\n\nJung's work is not the first proof of this result. Bence Hervay independently posted a computer-assisted proof on GitHub on September 28, a week before Jung's paper. An audit of Hervay's proof found no mathematical gaps but noted it lacked peer review. Oscar Smith's proof is formalized in the Lean proof assistant.\n\nJung acknowledges that his paper has not undergone peer review and that another paper reached the same conclusion around the same time through a different approach. Despite the mixed reactions from online commenters, Jung believes the conclusion is likely accurate, as he had personally verified his work. However, he emphasizes that the paper must still pass peer review to be considered an academic contribution.\n\nExperts at the Institute for Basic Science and Seoul National University have pointed out that while Jung's paper establishes a public record of its release, it does not guarantee the validity of its contents. They stress that the author must understand every aspect of the research and be able to answer questions from others for the work to be recognized as a complete academic achievement.",
  "summary": "A South Korean ninth grader says he used Anthropic's Claude to prove that a 62-faced solid cannot pass through a hole in an identical copy of itself, a claim not yet peer reviewed.",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}