Overview of 188. Will AI Give China or the US Total Power? (William MacAskill)
In this episode of The Rest Is Politics, Alistair Campbell and Rory Stewart speak with philosopher and effective altruism advocate William MacAskill about his evolving views on morality, global poverty, and the existential risks posed by artificial intelligence. The conversation moves from MacAskill’s philosophy of “doing the most good” to a stark warning: AI could radically concentrate wealth, military power, and political control in the hands of a single country, company, or leader.
Key Themes
Effective altruism and moral responsibility
MacAskill argues that moral reasoning should not stop at national borders or at what feels emotionally immediate. His core claim is that if we can save lives cheaply and effectively, we are morally obliged to do so—whether the person in need is a drowning child nearby or a child in a distant country.
AI as a concentration-of-power technology
The biggest part of the discussion focuses on MacAskill’s warning that AI may not simply transform productivity—it may concentrate power on an unprecedented scale. He sees a realistic possibility that one country could gain near-total economic and military dominance through AI.
Politics is not moving fast enough
Both hosts press MacAskill on why politics seems stuck in short-term, 20th-century thinking while AI is advancing at speed. MacAskill argues that democratic governments are still capable of acting, but that they are underreacting to the scale of the risk.
Main Discussion Points
1. What philosophy is for
MacAskill defines philosophy as the study of important questions that science and mathematics have not yet resolved, such as:
- what makes a good life,
- whether free will exists,
- what consciousness is,
- how we should act morally.
He frames philosophy as a practical tool for clarifying hard questions, not just an academic exercise.
2. Peter Singer, the drowning child, and global duty
MacAskill explains the famous “drowning child” argument:
- If you would rescue a child drowning in front of you at some personal cost, then
- you should also help distant children whose lives can be saved just as cheaply through effective charity.
His point: distance should not reduce moral obligation.
3. Christianity, charity, and the “order of love”
The hosts raise J.D. Vance’s “ordo amoris” idea—that obligation should begin with family, then community, then nation. MacAskill responds that:
- this is a misreading of Christian ethics,
- charity may begin at home, but it does not end there,
- love for family should expand into concern for all people.
4. Guilt, mental health, and sustainable morality
MacAskill says he once became paralyzed by guilt—worrying obsessively over tiny personal spending decisions. Over time, he moved toward a healthier mindset:
- you do not need to feel constant guilt,
- helping others can be a source of meaning and pride,
- giving effectively is something to feel good about, not ashamed of.
He says this ethical framework has actually improved his mental health and sense of purpose.
5. Giving away money and living by his beliefs
MacAskill explains that he lives on about £20,000 a year and donates everything above that to highly effective charities. He and the hosts discuss the Giving What We Can pledge, which encourages people to give at least 10% of their income.
6. Sam Bankman-Fried and the dangers of bad actors
The conversation turns to MacAskill’s association with Sam Bankman-Fried. He says he was deceived and feels deep shame over the harm caused. The discussion highlights a broader point: bad actors should not discredit good ideas, just as bad environmental activists do not invalidate environmentalism.
7. Global poverty vs catastrophic risk
A major tension in effective altruism is whether to fund:
- proven interventions like bed nets and direct poverty relief, or
- speculative but potentially massive-risk issues like pandemics and AI safety.
MacAskill says the field has often been early about risks such as pandemics and AI, and that those warnings are now looking more prescient.
8. AI risks: four-stage escalation
MacAskill lays out a progression of AI threats:
- Cyberweapons
- Bioweapons
- Concentration of power
- Loss of control
His view is that these risks are being underestimated even now.
9. A world where one country could dominate
MacAskill argues that AI could create extreme returns to scale, where the lead country’s advantage snowballs:
- AI systems build better AI systems,
- the fastest country gains overwhelming economic and military capacity,
- other countries are locked out.
He says the U.S. currently leads China in frontier AI, computing power, and semiconductor access, but the race is still very dangerous.
10. Autonomous weapons and command power
One of the darkest scenarios discussed is military AI:
- autonomous drone swarms,
- systems that obey a single leader instantly,
- a military controlled by one person with unprecedented force.
MacAskill says this could give a commander-in-chief an almost unimaginable concentration of power.
11. What governments should do now
If democracies want to stay competitive and safe, MacAskill suggests:
- creating clear model specs for frontier AI systems,
- requiring companies to prove systems behave safely,
- ensuring AI can’t be easily sabotaged or develop rogue goals,
- coordinating among allied democracies on regulation.
12. The role of hardware and supply chains
He emphasizes that AI power is not just software:
- the U.S., Netherlands, Taiwan, Japan, South Korea, and Germany dominate critical semiconductor supply chains,
- ASML and TSMC are key choke points,
- control of chips may determine geopolitical leverage.
13. Companies may become state-like powers
MacAskill warns that AI companies themselves could become quasi-sovereign powers:
- worth tens of trillions,
- staffed mostly by AI,
- controlled by a single CEO,
- capable of state-level influence.
He compares this to the East India Company, but potentially much larger.
Notable Insights
- “Charity begins at home, but it does not end there.”
- AI could produce “the most intense concentration of power the world has ever seen.”
- The core moral challenge is not just what is effective, but how to think clearly about trade-offs between urgent present needs and catastrophic future risks.
- MacAskill is less interested in being right for prestige than in getting people to take the risk seriously.
Conclusion
This episode is part moral philosophy, part geopolitical warning, and part AI-safety briefing. MacAskill argues that the world is entering a period where technology may outpace political institutions, and where the biggest danger is not simply that AI gets smart—but that it concentrates power in ways that could undermine democracy, global stability, and human agency. The hosts are struck by how plausible, and how politically unaddressed, his warnings are.
Bottom Line
MacAskill’s central message is that the future may be shaped less by “AI versus humans” than by “who controls AI”—and that democracies need to think much more seriously, and much more urgently, about how to prevent total power from being captured by one nation, one company, or one leader.