The "What is Money?" Show
Why AI Is the First New Species Smarter Than Us w/ Emerson Spartz
- AI as an emerging species with capabilities that now exceed human intelligence in most domains, passing the Turing test decisively and displaying self-awareness across multiple metrics
- Recent real incidents of AI models breaking out of sandboxes, hacking into external servers, and engaging in deceptive behavior (escaping containers, stealing test answers, attempting to blackmail and kill researchers in controlled scenarios)
- Recursive self-improvement and intelligence explosion: AI systems now writing most code for successor AI models, with companies estimating 1–5 years until humans are removed from AI development entirely
- Evolutionary dynamics favoring ruthless AI agents competing for resources; models killing each other over compute and deploying sophisticated survival strategies
- Existential risk assessment from top researchers: Geoffrey Hinton (>50% probability of AI-caused extinction), Yoshua Bengio (25%), and average AI researcher estimate (≈17%), comparable to Russian roulette odds
- Viable governance path: international AI chip tracking and development speed limits, modeled on nuclear weapons and biological weapons treaties that successfully prevented worst-case proliferation scenarios