Anthropic mathematician Levent Alpoge has claimed to discover a counterexample to the Jacobian conjecture, which has remained unsolved since 1939. According to him, the potential refutation was proposed with the help of the Claude Fable 5 model and represents a polynomial mapping in $\mathbb{C}^3$ with a constant Jacobian determinant equal to -2.
What Happened
Levent Alpoge (Anthropic) presented a result from the Claude Fable 5 model that indicates the existence of a counterexample to the Jacobian conjecture. The proposed mapping in $\mathbb{C}^3$ space has a constant Jacobian determinant of -2, yet the mapping itself is not invertible. At this time, this claim has not undergone official scientific peer review.
Context
The Jacobian conjecture is one of the fundamental open problems in algebraic geometry, remaining unsolved for decades. The use of frontier models, such as Claude Fable 5, to solve such problems marks a transition for AI from text generation to acting as an intellectual "scientific draft" in fundamental science.
Why It Matters for the Industry
For the AI industry, this event could be a turning point, demonstrating the ability of models to solve tasks of extremely high abstraction. This intensifies the competition between Anthropic, OpenAI, and Google in the areas of mathematical precision and reasoning, and paves the way for the creation of specialized AI agents for scientific discovery.
Why It Matters for Users
For researchers and scientists, this signifies the emergence of a new class of tools that can act as "research co-authors," helping to verify complex hypotheses and find mathematical anomalies that might be missed by humans. This shifts the paradigm of LLM usage from simple chatbots to deep R&D tools.
What Is Still Unknown / Limitations
The result has not yet been verified by the scientific community and requires rigorous peer review to confirm its mathematical correctness.
Sources
Author
Look at AI, Editorial Staff