This Is the Good Scenario?

Ezra Klein · The New York Times, Opinion · October 18, 2026 · nytimes.com (Klein’s gift link, from his tweet: append ?unlocked_article_code=1.KFE.2ZLx.eu5XGDgQoHCe). Summary below with a few short quotes; the full column is behind the link.

Summary

The thought experiment. Klein opens with a button: press it and China becomes the first hyperpower, leaving Americans with only a dim understanding of the frontier, but in return Americans become richer than ever, with cures, cheap energy and freedom from work. Cede control and receive abundance. Would you press it? His point is that this is roughly the success scenario AI’s builders describe, with AI in China’s role. The AI debate centres p(doom); he wants to examine p(utopia).

Coxon and Stewart. Jacob Coxon, the 27-year-old former OpenAI and Anthropic researcher, went on The Daily Show to warn about catastrophe. When Jon Stewart mocked the industry’s “future utopia,” Coxon said he believes in it, and described it: no jobs, a universal high income funded by the dividends of fully automated production. Stewart’s reply was whether we would all just be performing in a play. Klein says this vision is common inside AI companies and most people find it anything but utopian.

Mathematics. Klein is sceptical of a jobless future but finds it easy to imagine humans handing over control and comprehension. His first example: OpenAI released 700+ manuscripts proposing solutions to nearly 400 advanced math problems at once. His father, a retired mathematician, says two problems at the centre of his field, on electrons moving through disordered crystals, were “totally out of reach”; OpenAI’s systems solved them in an afternoon. A newly formed Association for Human Mathematics (about 1,000 members) responded that mathematicians “did not ask for this work to be done.” He also cites a study in which the best model solved 98% of 50 unpublished theoretical-physics problems at under $3 each, and corrected the physicists who thought it had erred.

Gradual disempowerment. The scenario he finds most likely is not a single moment of lost control but gradual disempowerment: firms, militaries and labs each hand more decisions to AI because competitors who do so win. Even if the AIs stay helpful, more of society’s workings slip beyond human understanding.

Meta’s Muse. His mundane example: Meta’s Muse assistant, at the top of the App Store for weeks, works because it has access to email, calendar, social accounts, bank accounts and cloud storage and acts on the user’s behalf. Friends say it makes them feel superhuman. But businesses on the other end are already swamped by automated calls and emails and will answer with their own agents, until it is agents bargaining with agents, and anyone without one falls behind. As with email and smartphones, opting out becomes too costly.

Fukuyama and Huxley. In The End of History and the Last Man, Fukuyama asked whether liberal democracy, unbeatable at producing prosperity, would betray the human drive for recognition and struggle. Klein argues AI poses the question more sharply: it offers prosperity at the price of human primacy, or in his phrase, it “promises inferiority as a condition of prosperity.” He quotes Stewart’s objection that people need to feel relevant, and that many people his age die soon after retiring for lack of purpose. Coxon answered that many people hate their jobs; Klein adds that people love retirement but loathe unemployment, and that the deeper loss is purpose and control rather than jobs. He closes with Brave New World, where every desire is met and talk of purpose is subversive, and notes that the anger of the politically radicalised rich shows how little wealth satisfies once it is cut off from recognition.

The close. He does not say AI will create this world, only that many of its builders want it, and each new result makes it more plausible. Quoting Stewart that the builders don’t seem to understand the human condition, he guesses almost all of them would press the button, and asks whether the reader would.

Connections

• Ezra Klein — author.
• Jacob Coxon on The Daily Show — the interview he builds the column around. The Most Revealing Part of This… (tweet) is his post sharing it.
• The End of History and the Last Man — Fukuyama’s thymos argument, applied to AI.
• Brave New World — the closing frame. Kyle has read it.
• Gradual Disempowerment — the mechanism he finds most likely.
• Universal Basic Income — Coxon’s “universal high income” is the strong form.
• Food Should Be Free… (tweet) — the meme Kyle saved the same day, mocking people who want abundance without automation; Klein is making the serious case for hesitation.
• Empires of AI — Ferguson, Klaassen and Schmidt, published two days later, make the same competitive-pressure argument at the level of nations instead of firms and people.