Скачать или смотреть Stuart Russell: AI CEOs are aware of extinction risks but continue the race.

#stuartrussell #agi #airesearch #aisafety #openai #existentialrisk

We sit down with Stuart Russell, a leading AI expert, Berkeley professor for 40 years, and co-author of the definitive AI textbook (written in 1992), to discuss the most pressing existential risks of Artificial General Intelligence (AGI). Russell reveals shocking private conversations with top AI CEOs, including one who suggested a Chernobyl-level disaster might be the "best-case scenario" needed to force government regulation before an irreversible loss of control.

Russell outlines the "race" dynamic, where CEOs (like those at OpenAI and DeepMind) acknowledge the extinction-level risk—comparing AGI to nuclear war and pandemics—but feel they cannot stop the development due to investor pressure and market forces. The colossal scale of investment, expected to reach a trillion dollars next year (50 times the budget of the Manhattan Project), underscores the speed and danger of this race.

He clarifies the distinction between AI and AGI, emphasizing that AGI doesn't require a robot body to exert immense power; its ability to influence 3/4 of the world's population via the internet and language makes it a global threat. Russell introduces the "Gorilla Problem" to explain the potential fate of humanity when faced with a species (AGI) far superior in intelligence. The core question remains: Can humans, driven by economic value and scientific seduction, ensure that superintelligent machines will always act in our best interest before we become the "gorillas"?