Can We Trust AI? The Future of Verified Reasoning in High-Stakes Systems
Thu Jan 22 2026
In this episode of Founded & Funded, Madrona Partner Jon Turow hosts a live conversation with Carina Hong, Founder & CEO of Axiom, and Byron Cook, VP & Distinguished Scientist at AWS.
Carina is building foundation models trained on verified proofs instead of human-written reasoning. Byron leads AWS's automated reasoning group, which secures massive infrastructure with mathematically proven systems.
During this live IA Summit conversation, they explore what it means to build AI that actually reasons and why verified intelligence is critical for domains where being right really matters.
They dive into:
1) Why today's models fail at objective truth in high-stakes domains
2) How verification bridges the gap between correctness and scale
3) What superintelligent reasoning engines might unlock
4) The future of AI in finance, chips, healthcare, and beyond
This is a must-watch for any founder or builder working in AI, infrastructure, or high-consequence systems.
Full Transcript: https://www.madrona.com/can-we-trust-ai-the-future-of-verified-reasoning-in-high-stakes-systems
Chapters:
(00:00:00) Introduction
(00:02:01) Meet Carina Hong & Byron Cook
(00:04:07) Why Next-Gen Reasoning Models?
(00:05:59) Objective Truth & Verification in AI
(00:08:03) Formal Verification at Amazon
(00:09:40) Making Proof Tools Usable
(00:11:14) Proofs vs. Bugs: The Mathematical Approach
(00:13:25) The Market for Reasoning & Scarcity
(00:15:41) From Scarcity to Abundance in Reasoning
(00:17:47) AI Mathematicians & Scientific Breakthroughs
(00:19:59) Collaboration: AI & Human Experts
(00:22:16) Lowering the Cost of Creativity & Experimentation
(00:24:08) Broad Applications of Mathematical Reasoning
(00:25:28) Balancing Theory & Practice in AI
(00:28:14) Customer-Driven Investment in Formal Methods
(00:30:31) Building Toward Superintelligent Reasoning Engines
More
In this episode of Founded & Funded, Madrona Partner Jon Turow hosts a live conversation with Carina Hong, Founder & CEO of Axiom, and Byron Cook, VP & Distinguished Scientist at AWS. Carina is building foundation models trained on verified proofs instead of human-written reasoning. Byron leads AWS's automated reasoning group, which secures massive infrastructure with mathematically proven systems. During this live IA Summit conversation, they explore what it means to build AI that actually reasons and why verified intelligence is critical for domains where being right really matters. They dive into: 1) Why today's models fail at objective truth in high-stakes domains 2) How verification bridges the gap between correctness and scale 3) What superintelligent reasoning engines might unlock 4) The future of AI in finance, chips, healthcare, and beyond This is a must-watch for any founder or builder working in AI, infrastructure, or high-consequence systems. Full Transcript: https://www.madrona.com/can-we-trust-ai-the-future-of-verified-reasoning-in-high-stakes-systems Chapters: (00:00:00) Introduction (00:02:01) Meet Carina Hong & Byron Cook (00:04:07) Why Next-Gen Reasoning Models? (00:05:59) Objective Truth & Verification in AI (00:08:03) Formal Verification at Amazon (00:09:40) Making Proof Tools Usable (00:11:14) Proofs vs. Bugs: The Mathematical Approach (00:13:25) The Market for Reasoning & Scarcity (00:15:41) From Scarcity to Abundance in Reasoning (00:17:47) AI Mathematicians & Scientific Breakthroughs (00:19:59) Collaboration: AI & Human Experts (00:22:16) Lowering the Cost of Creativity & Experimentation (00:24:08) Broad Applications of Mathematical Reasoning (00:25:28) Balancing Theory & Practice in AI (00:28:14) Customer-Driven Investment in Formal Methods (00:30:31) Building Toward Superintelligent Reasoning Engines