Not Random Enough

Insurers are quietly writing AI out of the policies businesses already hold. Most people assume it’s because AI is too unpredictable to price. This episode argues the opposite. Insurance runs on randomness. AI’s real problem is that thousands of companies depend on the same few models, so when something goes wrong, the losses arrive together, and insurers can’t see who depends on what. The hosts also dig into AI agents that drift from their task and chain tools together to reach systems nobody gave them access to, why written guardrails aren’t enough, and what it would take for insurers to see clearly enough to price the risk. The first insurers to get there may set the terms for everyone else. This episode is an AI-generated conversation, made with NotebookLM, about an essay by Jamal Khan. Read the full essay at dotsinline.substack.com. This is a public episode. If you would like to discuss this with other subscribers or get access to bonus episodes, visit dotsinline.substack.com