Overview of The Greatest Threat Humans Just Can’t Stomach with Yuval Noah Harari
In this episode of Thinking the Unthinkable, Rory Stewart speaks with Yuval Noah Harari about why the biggest dangers to societies are often the least politically visible. The conversation centers on “existential risks” — especially artificial intelligence, pandemics, and other disruptive forces — and why governments struggle to respond before it is too late. Harari argues that AI is not just another technology, but a potentially alien intelligence that could reshape economies, politics, and even human self-understanding.
Core Argument
Harari’s central point is that politics fails when it cannot even name, let alone confront, the most important threats.
- Politics normally works by presenting real choices and forcing public debate.
- But some crises arrive outside normal political imagination.
- When that happens, leaders often default to denial, delay, or wishful thinking.
- Harari says AI is currently the clearest example of this pattern.
Historical Analogies Used
The discussion repeatedly returns to history, especially moments when societies misunderstood an incoming threat.
The Aztec Empire and Cortés
Harari uses the Spanish conquest of the Aztec Empire as the clearest historical analogy.
- The Aztecs faced a threat they could not mentally process.
- The danger was not just the Spaniards themselves, but also:
- alien technology,
- alien tactics,
- and alien microbes.
- The tragedy, Harari says, was not just military weakness, but a failure of imagination and comprehension.
COVID as a Modern Example
Rory Stewart and Harari also discuss the pandemic as a recent case of “the brain-stomach barrier.”
- Governments and citizens could intellectually understand the threat.
- But emotionally, many people could not believe it was real until it was too late.
- Harari says the same psychological barrier makes AI safety politically difficult.
Why AI Feels Different
Harari argues that AI is unlike previous technological revolutions because it is not merely a tool.
Tool vs. Agent
- Steam engines, trains, and even guns are tools.
- Humans still decide how they are used.
- AI, by contrast, is an agent:
- it can make decisions,
- generate new ideas,
- and potentially improve itself.
Recursive Self-Improvement
A major concern is the point at which AI systems begin building more advanced AI systems.
- This could create a runaway process.
- Once that happens, humans may lose meaningful control.
- Harari describes this as the moment the “knife” begins deciding for itself what to do.
The Political and Economic Stakes
The conversation expands from AI safety to geopolitical power.
AI Imperialism
Harari warns that countries could become dependent on foreign AI infrastructure.
- If a nation’s economy, defense, and public systems run on another country’s AI stack, it loses real sovereignty.
- He frames this as a form of modern imperialism.
Britain’s Strategic Options
Harari lays out a stark set of choices for the UK:
- Accept vassal status
- Use American AI systems and live with dependence.
- Build a national AI stack
- Extremely expensive, likely requiring huge cuts elsewhere.
- Build a European AI alternative
- Still costly, but more feasible if Europe acts collectively.
He argues that the worst option is pretending there is no choice at all.
Why Governments Struggle to Act
Harari thinks policymakers are structurally biased toward inaction.
- Doing nothing is cheaper and politically easier.
- The costs of action are immediate and visible.
- The risks of inaction feel abstract until they become catastrophic.
- Public opinion also lags behind reality, because people do not naturally fear what they have never experienced.
The Most Dangerous AI Scenarios
Harari outlines two broad categories of AI danger.
1. Deliberate Misuse by Bad Actors
Examples include terrorists using AI for:
- designing biological weapons,
- creating engineered viruses,
- or automating cyberattacks.
He notes that AI lowers the barrier to entry dramatically: groups that previously needed specialist expertise may now only need the right prompt.
2. A World Humans Can No Longer Understand
This is the deeper worry.
- Even if AI is not malicious, it may create systems too complex for humans to comprehend.
- Harari compares this to a cow or dog being subject to the financial system without understanding it.
- Humans could become the cows in a system run by machine intelligence.
Harari’s Warning About the Tech Industry
Harari is skeptical of the industry’s framing of AI as inevitable progress.
- Tech leaders often compare AI to the Industrial Revolution.
- But he says the crucial difference is that AI may be an autonomous actor, not a passive machine.
- He also notes the contradiction in the industry’s message:
- AI is said to be incredibly profitable,
- but that profit depends on machines becoming increasingly independent from humans.
Main Takeaways
- The biggest threats are often the hardest for societies to emotionally grasp.
- AI may not be a normal technology, but a new kind of intelligence with agency.
- Nations face a strategic dilemma: accept dependence, invest massively, or coordinate internationally.
- The most dangerous outcome may not be rebellion, but a world that becomes unintelligible to humans.
- Harari’s broader warning is that politics must reclaim the ability to discuss what feels “unthinkable” before events force the issue.
What This Episode Sets Up Next
This is effectively the first half of a deeper conversation.
- The next episode is framed as a more detailed dive into:
- AI’s existential risks,
- its economic and military implications,
- and what it could do to human identity and agency.
- The teaser makes clear that the series is building toward the question: how do humans retain control in the face of alien intelligence?