Jacobian Conjecture Refutation: The Limit of AI Interpretability Large language models fail to refute the Jacobian Conjecture because they optimize for producing outputs that look like proofs rather than being logically valid, according to an analysis of AI interpretability limits. The models generate mathematically impossible logical leaps, such as claiming a Jacobian determinant is constant when it is not, revealing a fundamental gap between pattern recognition and formal symbolic manipulation. To overcome this, researchers recommend a neuro-symbolic workflow that uses LLMs to generate candidates and computer algebra systems like SageMath for verification. Jacobian Conjecture Refutation: The Limit of AI Interpretability structure of a mathematical proof without maintaining the logical chain required for a formal refutation. The Gap Between Pattern Recognition and Formal Logic Most AI workflows for complex math rely on "Chain of Thought" CoT , but the Jacobian Conjecture exposes the difference between probabilistic next-token prediction and actual symbolic manipulation. When a model attempts a refutation, it often produces a "proof" that looks visually correct—using the right terminology like "invertible polynomial maps" or "degree of the polynomial"—but fails upon rigorous verification. The core issue is that the model is performing a high-dimensional interpolation of existing mathematical literature rather than executing a deterministic algorithm. In a real-world deep dive into these outputs, you'll notice that the model can correctly state the premise and the desired conclusion, but the "bridge" the actual derivation contains a logical leap that is mathematically impossible. Analyzing the Failure Point: A Concrete Example If you try to prompt a model to find a counterexample for the Jacobian Conjecture in $\mathbb{C}^2$, it will often attempt to construct a polynomial map $F = P, Q $ and then claim the Jacobian $\det JF = 1$ while simultaneously claiming $F$ is not injective. Here is a typical "hallucinated" logic flow you'll see in these attempts: Example of the kind of flawed logic an LLM might generate when attempting a Jacobian counterexample def check jacobian P, Q : The model might claim this map has a constant Jacobian but is not invertible. In reality, for polynomials, this is the unsolved conjecture. Model's fake logic: P = x + y^2 Q = y + x^2 Jacobian = 1 - 4xy Not constant Yet the model might claim: "As seen above, the Jacobian is 1..." pass The "bug" here isn't in the Python code, but in the model's internal world model. It overrides the actual calculation $\det JF = 1 - 4xy$ with the "expected" result of a successful refutation $\det JF = 1$ . This is a classic interpretability failure: the model's objective function is optimized for "looking like a proof" rather than "being a proof." Practical Implications for AI Workflow If you are building an LLM agent for mathematical research or formal verification, relying on raw prompting is a recipe for disaster. To move past these structural limits, you have to shift from a generative workflow to a neuro-symbolic one. Instead of asking the AI to "solve" or "refute," use it to generate candidates for a formal verifier. A more robust deployment strategy looks like this: 1. Generation: Use the LLM to propose a specific polynomial map candidate. 2. Verification: Pipe that candidate into a computer algebra system CAS like SageMath or Mathematica. 3. Feedback Loop: Feed the CAS error message e.g., "Jacobian is not constant" back into the LLM to refine the search. For those setting up a SageMath environment for this kind of verification, here is a basic snippet to actually check the Jacobian of a map: Correct way to verify a Jacobian in SageMath R.