What was already known
Erdős Problem 769 asked for bounds on the eventual cube-tiling count c(n), especially whether c(n)≫n^n. The cited Star Fleet account does not establish a sharper pre-solution quantitative bound.
AI involvement
Star Fleet presents the tiling construction as work of autonomous GPT-5.6 proof agents. Colin Snyder built the system and selected tasks; the displayed referee is another AI agent, and no named human checker of this proof is documented.
The first-person system account assigns separate problems to autonomous GPT-5.6 agents and reports no human mathematical input to this construction beyond setting and reviewing the task.
Importance
Refutes the proposed universal c(n)≫n^n lower bound along odd dimensions.
Limitations
The refutation uses odd dimensions n≥201 and does not determine sharp c(n) bounds in all dimensions.
Technical verification
The source reports a Lean 4 theorem negating the stated lower-bound proposition, a scope audit of the formal definition, a clean source rebuild and a second fresh-machine build. These are source-reported checks. The referee is an AI agent, so no human method is recorded.
Formal checking
Pinned verifier, clean Lean build, axiom and placeholder scan, plus fresh GCP source rebuild
- Checker
- Lean 4 kernel, reported by Star Fleet and its Claude referee
- Scope
- Erdos769.erdos769_lower_bound_false: the canonical c(n)≫n^n proposition is false using explicit odd-dimensional tilings
- Basis
- Reported by the linked checker
Checking a formal statement does not establish that it matches the original problem.
Human understanding
The source gives an agent referee's explanation of the grid, substitution and conductor chain but no named person's explanation of the central construction. Few is the documentation default.
Correction and publication history
Refutes the proposed universal c(n)≫n^n lower bound along odd dimensions.
Read this published revision