She Raised $64M to Build an AI Math Prodigy | Carina Hong, CEO of Axiom
Summary
Axim is developing a self-improving reasoning engine that integrates both generation and verification capabilities, utilizing formal languages like Lean to ground natural language processing. This approach combines deterministic tooling with probabilistic systems. The company recently achieved a score of 8 out of 12 on the Putnam Mathematical Competition, a notoriously difficult contest for math undergraduates where the median score is typically zero. Carina Hong, Axim's founder, transitioned from an academic path to entrepreneurship, driven by the "contrarian appeal" and the adventure of building a new company. She envisions a future where mathematicians collaborate with AI, using AI as a "diligent grad student" to prove their intuitions.
Key takeaway
For research scientists exploring AI applications in mathematics, you should consider how integrating formal languages with probabilistic AI systems can enhance reasoning and verification. This approach allows AI to function as a powerful tool for proving complex intuitions, potentially shifting your focus to higher-level abstraction and intuition-driven problem-solving.
Key insights
Axim combines formal languages and AI to create a self-improving reasoning engine for mathematical verification.
Principles
- Combine deterministic and probabilistic systems.
- AI can serve as a diligent assistant for human intuition.
Method
Axim's reasoning engine uses formal languages (e.g., Lean) to ground natural language, integrating generation and verification to achieve self-improvement.
In practice
- Use formal languages for AI grounding.
- Apply AI for mathematical proof verification.
Topics
- Reasoning Engines
- AI in Mathematics
- Formal Languages
- AI Verification
- Probabilistic Systems
Best for: Research Scientist, AI Researcher, AI Scientist, Entrepreneur
Related on AIssential
Editorial summary, takeaway, and curation by AIssential. Original article published by Weights & Biases.