Just now, GPT-6 Astra teamed up with a non-professional mathematician to prove a 57-year-old "circle packing" problem.

What happened: A non-professional mathematician using GPT-6 Astra proved the remaining cases of a 57-year-old circle packing problem for N=15, 16, 17, 18 and 20. Key numbers: N up to 20; N≤14 and N=19 were previously verified; N≤7 proven in 1968; N≤10 in 1969; N≤14 and N=19 around 2003. Why it matters: The result extends verified optimality beyond the earlier range, and the author says AI-assisted work on the same problem drew interest from other researchers online.

GPT-6 Astra and a non-professional mathematician proved a 57-year-old circle packing problem.

GPT-6 Astra and a non-professional mathematician proved a 57-year-old circle packing problem. — impact context not separately stored; see What changed.