Steven Bartlett with Yoshua Bengio
Caring for his grandson forced an AI pioneer to confront a question he had pushed aside: what kind of world would his research help leave behind?
Caring for his young grandson served as Bengio's personal turning point, shifting his focus toward AI safety after realizing the catastrophic risks facing future generations.
Although the podcast title references a two-year threshold, Bengio does not offer a precise personal forecast; he describes uncertain capability timelines from a few years to a decade or two while treating probabilities of harm as a separate question.
Under the precautionary principle, Bengio argues that even a minor percentage chance of existential threat or democratic collapse makes inaction morally unacceptable.