The Greatest Threat Humans Just Can't Stomach | with Yuval Noah Harari

Summary

Yuval Noah Harari argues that politics must confront threats that fall outside familiar ways of thinking. He uses the Spanish conquest of the Aztec Empire as a historical analogy: the Aztecs had the numbers to resist Cortés’s small force, but struggled to understand what they were facing. Harari says societies today risk a similar failure to grasp the scale and nature of AI.

He describes AI not as an ordinary tool but as an “alien” form of inorganic intelligence. Unlike a steam engine or a knife, an AI agent can make decisions and generate new ideas; future systems may also help build more capable AI. Harari warns that this could make events move faster than human institutions can respond. The discussion links this to the early COVID response: people and governments may understand a danger intellectually while instinctively dismissing it as too strange or costly to act on.

The conversation also examines the UK’s dependence on foreign AI infrastructure. Harari sets out three broad choices:

• Accept dependence on the United States and act accordingly.
• Attempt to build a sovereign British AI system, which he estimates could cost around £500 billion.
• Develop AI jointly with European countries, requiring much closer integration.

He also raises open-source development as a possible alternative to reliance on closed US systems. The speakers stress that each option has costs and risks, but argue that failing to choose is itself a consequential decision.

They challenge the claim that an AI race is inevitable. Harari says companies’ assertions that no country can afford to slow down or opt out can crowd politics out of the debate, even though governments still have choices. At the same time, the speakers acknowledge that refusing to develop AI while others press ahead could leave a country economically and militarily dependent.

The risks discussed range from malicious uses—such as AI helping people without specialist expertise design dangerous biological agents—to a broader loss of human control. Harari’s concern is not only that AI might deliberately harm humanity, but that it could create systems, including financial ones, that are too complex for people to understand or govern. The episode closes by framing AI as a threat to human independence and self-understanding, as well as a geopolitical and economic challenge.

Presenters and contributors

Rory Stewart; Yuval Noah Harari.