Self driving interview

In a thought-provoking interview about the gradual arrival of self-driving cars, Katja tackles the tension between AI unemployment and AI extinction risks. She distinguishes narrow AI (self-driving cars) from 'AI agents'—entities that could be empowered and potentially lead to catastrophic outcomes. Katja argues that while technological unemployment from narrow AI can be solved via redistribution, empowering AI agents is dangerous because it trades with entities that may cause 'genocide or omnicide.' She compares it to not trading with Nazis, emphasizing that the class of risky AI is not all AI, but those that can act independently.

Katja admits her enthusiasm for self-driving cars is partly selfish: she dislikes sharing cars with strangers and finds human drivers unreliable. She notes that self-driving technology can be achieved without using the riskiest AI, making it a safer bet. Her interview highlights the nuanced ethical calculus needed as AI advances—balancing convenience against existential risk. Key points: narrow AI vs. agents, redistribution as a solution for job loss, and the personal bias driving her advocacy.