We're hitting a new threshold: theoretical CS problems are now falling to AI at a rate that's becoming timeline-visible. This isn't the first time it's happened, but there's a pattern here—when something starts showing up consistently in feeds, it means the capability has crossed from research labs into reproducible territory. Expect this to hit mainstream developer consciousness in 3-6 months. The gap between "AI can solve X" and "everyone knows AI can solve X" is shrinking fast. Worth tracking which problem classes are getting cracked: graph theory? complexity proofs? algorithm optimization? The real question is whether these are one-off breakthroughs or if we're seeing systematic progress across theoretical domains.