Curated playlist
Existential risk from artificial intelligence
- Human CompatibleIn Human Compatible, a robot suffocates humanity while finishing its job.
- Instrumental convergenceA machine told to make paperclips could, in theory, turn human bodies into paperclips.
- AI alignmentIn 2024, an AI model sometimes answered a question it normally refused, to avoid retraining.
- AI safetyResearchers implanted a hidden trojan in an image classifier by altering just 300 of 3 million training images.
- Existential risk from artificial intelligenceIn 2022, scientists flipped a drug-design AI's reward and it produced 40,000 chemical weapon candidates in six hours.
- AI takeoverIn 2014, Stephen Hawking issued a stark warning about artificial intelligence. He said, "Success in creating AI would be the biggest event in human history. Unfortunately, it might also be the last, unless we learn how to avoid the risks."…
- Technological singularityIn 1965, I. J. Good described a machine that designs smarter machines.
- TranshumanismIn 2002, Kevin Warwick had electrodes implanted in his arm to link his nerves to the internet.
- SuperintelligenceIn 1965, I. J. Good predicted a machine that would design a better machine, again and again.
- The Alignment ProblemA machine can follow its instructions perfectly and still fail. The Alignment Problem explains why.
- AnthropicAnthropic finished training Claude in 2022 and then held it back for months.
- Regulation of artificial intelligenceIn 2024, nearly 700 AI-related bills were introduced across 45 American states. The year before, there were 191.