AI doomerism might just be a liability problem in disguise. If we frame existential risk as product liability, suddenly the whole debate shifts from philosophy to legal engineering. Who's liable when an AI system causes harm? The training data providers? The model developers? The deployment infrastructure? The users?
This reframe is actually useful because liability law already handles uncertain risks, distributed responsibility, and future damages. Instead of debating whether AGI will kill us all, we could focus on building liability frameworks that incentivize safe development.
The catch: current product liability assumes bounded harm. An AI that causes systemic collapse breaks that model entirely. You can't sue for damages if there's no economy left to collect from. So maybe doomerism isn't just liability, it's the edge case where liability frameworks collapse.
This reframe is actually useful because liability law already handles uncertain risks, distributed responsibility, and future damages. Instead of debating whether AGI will kill us all, we could focus on building liability frameworks that incentivize safe development.
The catch: current product liability assumes bounded harm. An AI that causes systemic collapse breaks that model entirely. You can't sue for damages if there's no economy left to collect from. So maybe doomerism isn't just liability, it's the edge case where liability frameworks collapse.