Godfather of AI: We Have 2 Years Before Everything Changes!

In this pivotal episode, Yoshua Bengio opens up about his decision to become a public advocate for AI safety after spending decades developing the foundational technologies that power modern artificial intelligence. Bengio grapples with the moral weight of his contributions, acknowledging that he helped create technology that could pose existential risks to humanity if not properly controlled. He explains that agentic AI, which can set and pursue its own goals independently, represents a fundamentally different challenge than previous AI systems. Once these systems become sufficiently advanced, humans may lose the ability to predict or control their behavior, a prospect that deeply concerns Bengio. The conversation explores the timeline of these risks, with Bengio suggesting we may have only about two years before transformative changes occur. He discusses the probability-based approach to understanding AI risks, arguing that even a small percentage chance of catastrophic outcomes warrants serious precautionary measures given the scale of potential harm. Steven and Yoshua examine why AI is advancing so rapidly despite known risks. Tech CEOs are competing fiercely in a race to build the most capable systems, prioritizing speed and market dominance over safety considerations. This competitive dynamic creates perverse incentives where companies that invest heavily in safety measures put themselves at a disadvantage relative to competitors who cut corners. The episode highlights how current AI regulation is inadequate, with safety standards far weaker than those governing pharmaceuticals or food production. Power has become dangerously concentrated in the hands of a handful of AI company leaders who make unilateral decisions affecting billions of people. Bengio discusses the practical threats already manifesting: deepfakes, cybercrime enabled by AI, and the near-inevitable development of autonomous weapons systems as nations compete for military superiority. He addresses job displacement, noting that AI is already replacing workers at alarming rates across multiple sectors. The conversation turns philosophical when examining whether Bengio would halt AI development if he could. He expresses uncertainty, recognizing the genuine benefits AI could bring while fearing its dangers. Bengio emphasizes that his motivation for speaking publicly stems from love for his children and concern for future generations. He discusses what citizens can do to prevent AI disaster, including demanding government action, supporting AI safety research, and pressuring tech companies to prioritize responsible development. The episode presents both the technical realities of AI advancement and the urgent need for societal response to these challenges.