Results for “Carina Hong”

There's a compelling mission—both because of the enormous economic opportunity it represents, if it succeeds, and because of the broader impact it can have on the world. The mission is defined by a stubborn technical challenge. It has to be hard. No one wants to work on something that isn't difficult, and in a sense, choosing the hardest problems becomes a moat.

Everyone on the team strongly believes in the mission. This is the one thing we want to work on in our lives, and multiple people share that same sentiment. We've fallen in love with a very stubborn technical challenge. Convincing others is always hard—it requires conviction on both sides. What we've demonstrated is small-team speed: speed in hiring, speed in getting strong results, and speed in refining our intellectual understanding.

A model that's really strong at mathematical reasoning is likely to be strong at coding. And a model that's excellent at both math and code is often very good at analysing the nuts and bolts of legal reasoning as well. The third and deepest reason this matters is the ability to bridge different levels of abstraction. All of these domains involve multiple layers of abstraction, and the ability to move fluidly between those layers is likely to be extremely commercially valuable.

At Axiom we've raised $64 million—we're a small startup—and we recently won the Putnam competition. We scored 90 out of 120, which would have placed us above all ~4,000 human contestants last year and at the level of a Putnam Fellow, meaning top five in the world. If you tried to achieve that purely through informal methods—where hallucination is a persistent risk—getting to the same level of consistent correctness would likely require a lot more resources.

We see math as code and code as math. The real magic, and the key transition, comes from combining AI, programming languages, and mathematics—bringing all three pillars together. What we envision is humans using informal reasoning and intuition as a powerful guide, with formal systems then verifying those ideas. That interplay across layers is, I think, the real magic of combining multiple levels of abstraction.

If you tried to achieve that purely through informal methods—where hallucination is a persistent risk—getting to the same level of consistent correctness would likely require a lot more research effort and resources.

The third and deepest reason this matters—why it's not just commercially meaningful but potentially world-changing—is the ability to bridge different levels of abstraction.

A model that's really strong at mathematical reasoning is likely to be strong at coding. And a model that's excellent at both math and code is often very good at analysing the nuts and bolts of legal reasoning as well.

What we envision is humans using informal reasoning and intuition as a powerful guide, with formal systems then verifying those ideas. In this way, the formal system grounds high-level intuition.

We see math as code and code as math. The real magic, and the key transition, comes from combining AI, programming languages, and mathematics—bringing all three pillars together.