Taking AI Doom Seriously For 62 Minutes

0:00 Intro 1:50 Intelligence is broad ability 3:55 You don't need to be conscious to act toward a goal 6:18 Beyond human intelligence 8:44 A human with computer powers 18:45 The AI persuasion thing 19:33 Power seeking and self preservation 22:17 Value communication 24:27 Value formation 27:44 Theoretically building a human-level machine 30:23 Neural networks 34:39 Reinforcement learning 37:45 Reasoning models 41:29 Multi modal models 43:40 Agentic models 45:49 Improvement prospects 47:26 Interpretability 49:26 Doom probabilities 53:40 Weighing the upsides and downsides 57:10 Places to continue learning