Welcome to The Closing Argument, our verdict on the news, plus everything The Argument published and appeared in this week.
The Verdict, by Jerusalem Demsas
The apocalypse is a trump card. When the threat of human extinction looms, every other disagreement, concern, worry, or tradeoff is a distraction. When everything is at stake, nothing else matters.
The core problem of human civilization is that everyone disagrees with everyone else. If, like ants or bees, we could all simply agree on the highest goals and the best means, then we could finally move in lockstep, allocating resources to public health without consternation over whether they should have gone to education instead. Ordinary politics — from appropriations fights to lawsuits and tedious debates — is the cooperative technology we’ve built for living alongside people whose priorities we find wrong and offensive.
The apocalypse seduces us with the hope that we can finally put all that aside. “The lure of such generalized disaster as a fantasy is that it releases one from normal obligations,” critic Susan Sontag wrote about end-of-the-world movies in her essay “The Imagination of Disaster.” Normal obligations for an individual can mean exercising, cleaning your house, or disciplining your children. Normal obligations for a political system mean things like respecting civil rights, constraining the power of the executive, and following the law.
“A great enough disaster cancels all enmities and calls upon the utmost concentration of earth resources,” Sontag went on. At a time when our brains are under constant assault by the distraction of our smartphones and the 24/7 news cycle we can, finally, relax. There is just one thing that matters. (Finally!)
Earlier this week, Jacob Coxon, a machine-learning scientist at Anthropic, resigned and went viral when he tweeted: “The people building AI earnestly believe that it could kill us all by the end of the decade.” His former colleague added publicly, “I personally think [human extinction] > 10% within the next decade.”
I too am very worried about AI safety for many reasons: I think that we could turn over our weapons systems to misaligned AIs that go on to make immoral decisions, that hackers and hostile state actors could attack our infrastructure, steal billions of dollars, and undermine the international banking system, plunging us into a worldwide depression. I also worry that AI-enabled surveillance will spell the end of true freedom, that terrorists will be able to use AI to manufacture biological threats, and that we will turn over more and more of our brains and thinking and research and ingenuity to machines.
I am not worried that AI will kill every single human being alive (if only because there are more than 8 billion of us and we have proven quite adaptable), though anything is possible. Yet this outcome tends to be the overwhelming, explicit focus of prominent AI safety activists.
Apocalyptic thinking tempts us to trade hard-won rights for security — or simply to just do something, even if the actions we take are not efficacious.
Most recent end-times talk has come from climate activists who are largely correct to be concerned about the threat of carbon emissions. But what has really been accomplished by all this green doomerism? Notably, the single biggest reduction in U.S. emissions came from the shale revolution, which shifted electricity generation from coal to natural gas, rather than anything the climate movement did. My point is not that we should expect existential problems to solve themselves — famously, climate change is not yet solved; my point is that dooming does not reliably produce the habits of mind that yield effective public policy.
Doomers also have a habit of making bad tradeoffs: Environmentalist James Lovelock argued in 2010 that climate change might make it necessary to “put democracy on hold for a while.” Philosopher Nick Bostrom, a leading thinker in the AI space, argued that we might need universal, continuous, worldwide surveillance to prevent the emergence of catastrophic technologies. Following the Fukushima disaster in Japan, Germany wound down its nuclear energy program, increasing reliance on coal. In New York, then-Governor Andrew Cuomo called the Indian Point nuclear plant a “ticking time bomb,” shutting down emissions-free power generation at a terrible time for New Yorkers’ energy cost concerns and the climate.
When the stakes are existential, normal concerns appear grotesque: If the most prominent AI safety advocate argues for launching airstrikes at data centers, who will listen to the bespectacled lawyers asking whether such a plan complies with international law? In an emergency, no one listens to the proceduralists. Not when we’re marching Japanese-Americans into internment camps, and not when we’re wiretapping American citizens.
Of course I believe that extreme risks deserve to become top priorities. I am glad that nuclear nonproliferation became a leading issue; it spurred humanity to levels of international cooperation hitherto unseen. I hope the same happens for AI safety. But I’m worried that all the people arguing about the latter are getting too hung up on confusing metrics like p(doom), which seemingly require buying into a fringe ideology. Maybe I’m wrong and AI does threaten the human race itself, but apocalyptic thinking is likely to make it harder, not easier, to consider the tradeoffs of our available responses and determine whether our proposed actions will actually help. The threat of extinction cannot settle those questions for us. For that, we still have to do politics.
Midterm update
Lakshya Jain, our director of political data, uses data from The Argument’s polling operation to inform the Split Ticket 2026 House and Senate models. These models update every day; bookmark this page to see the odds of each party taking the House and the Senate.
Top stories this week, by Milan Singh
As you may have heard, AIs have been going rogue and hacking things they aren’t supposed to hack. This is not great. On Tuesday, Kelsey Piper wrote about how losing control of AI is the course being taken by major AI labs — even though these labs are very worried about what autonomous AIs do. This seems like a bad plan!
Continuing our anniversary “How do we live with each other?” series, Vox writer-turned-grantmaker Dylan Matthews wrote about how he came around on Big Philanthropy.
Also in the anniversary series, Emily Hamilton of the Mercatus Center and Yonah Freemark of the Urban Institute have a piece up digging into why children aren’t walking to school anymore. Over the last hundred years, schools have been built on bigger and bigger lots and in less walkable areas.
🌟Abundance Wins of the Week🌟
Reps. Scott Peters, D-Calif., and Rep. Gabe Evans, R-Colo., introduced a bipartisan bill to speed up permitting for transmission projects by holding them to the same standard that already exists for hydropower and natural gas pipelines.
NIMBYs on the New York City Council tried to block Mayor Zohran Mamdani’s pick for chair of the Board of Standards and Appeals — a key housing post. Why? Because the nominee, John Mangin, was a key architect of the 2024 City of Yes plan and because he helped draft successful housing-related ballot measures that limited the discretion of the City Council. Despite the opposition, Mangin prevailed and has been approved!
New Jersey Gov. Mikie Sherrill signed a law legalizing balcony solar units and cutting fees on such units, helping users lower their electricity bills.
Worth watching…
This week marked the 25th anniversary of 9/11. After the attacks, George W. Bush launched two wars, with the goal of not just catching the terrorists, but of spreading liberal democracy throughout the globe. On the podcast, Matt and Jerusalem talked about the decision-making that led al-Qaeda to attack the U.S. and the U.S. to go to war, along with what went wrong with the liberal internationalist vision.
Lakshya Jain and VoteHub’s Zachary Donnini were joined by Rep. Sarah McBride, D-Del., the first openly transgender member of Congress, on Substack Live. It’s been a rough few years for trans rights. Republicans have successfully used the sports and youth transition questions as wedge issues, and some Democratic candidates are now moving to the center on these issues as well. But McBride doesn’t believe that will make it harder for her to work with them if they’re elected.
What’s News with The Argument
The Argument recommends, by Milan Singh
By the power vested in me as the author of this section of the Sunday newsletter, I decreed that the topic for this week’s cultural recommendations was movies. (This is largely because I’ve been back home in Cambridge, Massachusetts, and have been watching a lot of movies.)
“Mad Max: Fury Road is the best action movie of the last 20 years,” said Jeremiah Johnson, who described it as a “nearly perfect thrill ride” combining “absurdly high-octane action with rich characters and world-building.”
Kobe Yank-Jacobs, my fellow Fellow, said that he rewatched The Graduate, noting that “it was shocking how rich it is, frame by frame.” He also praised the soundtrack, going so far as to say that “Simon & Garfunkel make the whole movie what it is.” If he had to summarize the film in one word, it would be “plastics.” “There is a great future in plastics,” he continued, using more than one word.
After being threatened with having their recommendations generated for them by Grok, the rest of the office chimed in as well.
Megha Fernandes watched Ratatouille for the very first time and called it a “perfect, and I mean perfect, movie.”
Our fearless and peerless leader, Jerusalem Demsas, “recently nostalgia-maxxed” by going to see Indiana Jones and the Last Crusade in theaters. In addition to the warm, fuzzy memories of watching Indiana Jones movies with her father when she was growing up, she also noted that it’s nice to enter “a cinematic world where all-American archaeologists beat up Nazis” for a few hours — as opposed to our present world, where members of our government are posting Nazi memes.
Justin Zuckerman, our video editor extraordinaire, rewatched Superbad after finding the DVD in a Little Free Blockbuster (which I didn’t know existed). He liked it a bit more when he was 13 but still found it very funny.
Maibritt Henkel watched About a Boy — a Hugh Grant movie featuring “a very young and very adorable Nicholas Hoult” — on a plane. She said it was “a pretty heartwarming film and worth watching even if only for the early 2000s aesthetic.”
Copy editor Eli Richman told me that the anniversary of 9/11 is a good opportunity to check out he pro shot of Come From Away on Apple TV: a musical (I’ll allow it) about the experiences of passengers whose planes were diverted to Newfoundland, Canada, after the U.S. shut down its airspace. While it’s “not an easy piece of media to get through without tears,” it does manage to be “funny and heartwarming at times” and has some great songs. “If you remember 9/11, it will hit you like a truck. And if you don't, it will give you a good sense of how scary it was in the moment.”
As for me, I saw four (4) movies this week. On Tuesdays, my local movie theater offers $5.75 tickets, so I went to see Cars — back in theaters for the 20th anniversary of the film — and Spider-Man: Brand New Day. Cars was great. I think it might be the best Pixar film of all time, and it’s nice to catch all the jokes written for adults that you missed as a kid. Spider-Man was solid — definitely better than what I expected for late-stage Marvel (read: slop) — but I almost wished that the movie hadn’t included the Hulk and Punisher and had focused on a more street-level villain like The Hand or Kingpin.
I also watched The Town (excellent; God Save Ben Affleck) and Green Book at home. I remember that the latter sparked a bit of discourse about white savior tropes when it first came out. I’m going to chalk that up to “woke 1 was crazy,” because I quite liked the movie and didn’t view it as problematic at all.
We have merch!
We have quarter-zips, keychains, hats, and stickers. Each one is a great conversation starter in its own way. Buy them here.
I think part of the reason for discussing AI's danger in terms of complete human extinction, vs. something more nuanced like "85% of humanity will die before we cure the engineered plague / rebuild the grid / cease the global nuclear war" is that inevitably some people are going to imagine themselves at part of the lucky whatever%, and then we're off to fantasizing about repopulating the Earth, cool postapocalyptic fashion, and looting vacant mcmansions.
Plus, if we're talking about some sort of true superintelligence, actual extinction becomes plausible. Legions of robot warriors destroying the combined militaries of the world is currently science fiction nonsense; killer machines exterminating out the last million surviving humans in the Amazon and on isolated islands after successive virus dispersals wiped out everyone else is more realistic, conceptually.
Please read ai-2027.com