Researchers Formalize 11 Elusive Turing Machines

Scientists at Google DeepMind and the University of Cambridge defined the behavior of 11 mysterious BB(6) machines.

Updated on Oct. 22, 2026 in Mathematics

Bold flat-color editorial illustration featuring eleven interconnected geometric logic gates, representing the formalization of complex Turing machine behavior. — Researchers at Google DeepMind and the University of Cambridge have successfully formalized the behavior rules for 11 complex six-state Turing machines. AI Illustration. Upload story photo >

Live Poll

Do you trust that artificial intelligence will help solve complex scientific problems more effectively than humans?

Researchers have successfully formalized the behavior rules for 11 complex six-state Turing machines, a set known as Cryptids. Using the Gemini 3.1 Pro AI system, the team analyzed candidate machines to determine their underlying logic.

Why it matters

This breakthrough provides deeper insight into the behavior of complex Turing machines within the Busy Beaver space. By applying formal verification methods, the team has expanded the understanding of computational limits.

The researchers examined 1,094 BB(6) holdouts, 65 BB(2,5) machines, and four BB(3,3) machines to identify 52 total candidates. Lean 4 software was used to verify the formal proofs of the behavior rules generated by the Gemini 3.1 Pro system.

The players

Google DeepMind

This subsidiary of Alphabet Inc. specializes in artificial intelligence and machine learning research.

University of Cambridge

Located in the United Kingdom, this public research university is one of the oldest and most prestigious academic institutions in the world.

The details

The team identified the 11 Cryptid machines through rigorous pattern detection and AI-generated conjectures. The resulting behavior rules were checked against formal proofs to ensure mathematical accuracy in the Lean 4 environment.

October 22, 2026: The research results were published.

The Big Picture

This study significantly advances the foundational efforts to solve the Busy Beaver problem by classifying previously unknown machine behaviors. The successful identification of Cryptids bridges a gap in existing computational theory regarding the limits of six-state machines.

While these findings are theoretical, the integration of AI like Gemini 3.1 Pro with formal verification tools like Lean 4 demonstrates a future where complex logical proofs are automated. This workflow may eventually assist in verifying code stability for critical infrastructure software.

The takeaway

The use of AI for mathematical proof verification represents a significant evolution in how researchers approach long-standing computational puzzles. Future efforts will likely continue to rely on automated conjectures to tackle other unsolved problems in formal logic.