How to Think About AI Risk

AI-risk arguments go sideways because two different debates share the same label.

Present-tense risks are documented and current: biased lending and hiring decisions, deepfake fraud, misinformation at scale, surveillance, workers displaced mid-career. These have case files — audits, lawsuits, receipts.

Long-term concerns ask what happens if systems become far more capable than us — whether we'd still be able to steer, correct, or shut them down. The evidence there is argument rather than incident report, and expert opinion genuinely spans 'urgent priority' to 'speculative'.