Uploaded February 2026 | Updated September 2026, 2 weeks ago
AI reasoning starts where verification is non-negotiable.
Carina Hong, Founder & CEO of Axiom, is building a self improving superintelligent reasoner by grounding AI in formal mathematics, a direction she describes as: “I feel like we see something that the world doesn’t see yet.”
Key ideas from the conversation:
• formal languages force models to confront every missing assumption
• deterministic systems paired with probabilistic models unlock new reliability
• AI should be treated as a collaborator, not a second class tool
• how a tool like Axiom can free experts to focus on intuition instead of verification
• how AI could reduce multi-year verification to weeks
As AI systems write more code, design more hardware, and make higher stakes decisions, verification stops being optional infrastructure.
AI reasoning starts where verification is non-negotiable.
Carina Hong, Founder & CEO of Axiom, is building a self improving superintelligent reasoner by grounding AI in formal mathematics, a direction she describes as: “I feel like we see something that the world doesn’t see yet.”
Key ideas from the conversation:
• formal languages force models to confront every missing assumption
• deterministic systems paired with probabilistic models unlock new reliability
• AI should be treated as a collaborator, not a second class tool
• how a tool like Axiom can free experts to focus on intuition instead of verification
• how AI could reduce multi-year verification to weeks
As AI systems write more code, design more hardware, and make higher stakes decisions, verification stops being optional infrastructure.
