Hey Frames of Space listeners,
We’ve just dropped a powerful new episode featuring journalist Dylan Matthews, and you won’t want to miss this conversation!
Dylan Matthews is a senior correspondent at Vox and one of the key writers behind Future Perfect, a section of the site dedicated to exploring ways to make the world better through evidence, reason, and moral philosophy. If you look at his body of work, you’ll find that he’s written about everything from AI risk to tax policy to foreign aid spending. And that eclecticism is part of what makes him so compelling.
Here’s a quick rundown of what you’ll find in this episode:
🎬 Dylan Defends The Last Jedi (Yes, Really) Dylan opens up about his affection for The Last Jedi, offering a take you might not expect. He shares what he loved about Rian Johnson’s direction and the movie’s themes—especially the choice to make Rey’s background refreshingly unspectacular.
🧐 Journalism, Nuance, and Negativity Bias Ever wondered why bad news feels so much more powerful than good news? Dylan gets technical about “negativity bias,” explaining why it warps both how news is covered and how we experience it. He points out that, especially among well-read audiences, cynicism can masquerade as sophistication—even when the truth is more complicated.
💵 The Impact of US Foreign Aid Cuts One of the biggest stories in this episode is the massive shift in US foreign aid policy. Dylan lays out the real-world consequences of recent cuts, particularly how the Trump administration’s changes to USAID have led to millions of unnecessary deaths. He gives historical context—did you know George W. Bush also saved millions with PEPFAR?—and why most Americans are barely aware of the scale or life-saving effects of these programs.
🤖 What About Artificial Intelligence? Dylan shares his views on AI, speculating on futures that could go dramatically right or wrong. While we could see exponential improvements for most people, he emphasizes that dramatic positive outcomes would require serious income redistribution and strong regulations to avoid catastrophic misuse. He’s particularly wary that current political trends are not moving in that direction.
💡 What Can You Do? Feeling motivated to help? Dylan recommends effective global health nonprofits like GiveWell’s top charities (Against Malaria Foundation, Malaria Consortium, Helen Keller International) for those wanting to counteract the worst effects of foreign aid cuts. Politically, he points to local and state elections as the most impactful places to get involved if you want to turn the tide against Trumpism.
Don’t miss out—listen to the full episode for all the details and more thought-provoking discussion.
Thanks so much for being part of our Frames of Space community. If you found this episode enlightening or inspiring, forward it to a friend or share it on social media! And as always, let us know what you think or what topics you’d like us to cover next.
Stay curious,
Andrew
Show Notes
"We’ll miss globalism when it’s gone" by Dylan Matthews, Vox
"What would a world without foreign aid look like?" by Dylan Matthews, Vox
You can find my Patreon page here.
Andrew Xu [00:00:00]:
Hello, everyone. My name is Andrew, and welcome back to Frames of Space. I am thrilled to share today's episode with you. But before we dive in, I've got a quick update. You see, starting with this week, I will be dialing things back and returning to the biweekly upload schedule that this podcast is known for. And I'll admit, it is a little bittersweet. I've loved being able to deliver content to all of you week after week, and I'm sure many of you have enjoyed that consistency too. That streak of weekly episodes was a blast, and I'm really grateful for the energy and support that it's provided during that time.
Andrew Xu [00:00:56]:
But here's the thing. While this podcast will be uploading episodes every other week for now, I am hopeful that sometime in the future, I will be able to return to weekly uploads. And that's where you come in. If you're not already aware, I have a Patreon account where you can support me and my work. I'll leave a link to it in the show notes. If I am able to gather up enough supporters over the next few months, it will make it much easier for me to get back to that weekly schedule sooner rather than later. So if you're able and interested, I would really appreciate your support. Every bit helps, and it means the world to me.
Andrew Xu [00:01:38]:
Now let's get on to the heart of today's episode. If there's one thing that I admire about Dylan Matthews, it's the fact that he doesn't really stick to one single beat in his coverage. Dylan Matthews is a senior correspondent at Vox and one of the key writers behind Future Perfect, which is a section of the site dedicated to exploring ways to make the world better through stuff like evidence and reason and moral philosophy. But if you look at his body of work, you'll find that he's written about everything from AI risk to tax policy to malaria prevention and much, much more. And I think that eclecticism is part of what makes him so compelling. In this episode, we discuss a lot of things, but at the heart of it is a question about how to write and think with nuance and precision. You see, Dylan has a real commitment to intellectual honesty and epistemic humility, even when it makes his writing more complicated or less satisfying from a rhetorical standpoint. And that same commitment is what shapes his perspective on journalism itself.
Andrew Xu [00:03:11]:
In our conversation, we talk about a trend known as negativity bias, which is the tendency for news media and audiences to fixate on what is going wrong rather than what's going right. Dylan explains how that bias shapes not just what stories get written, but also how educated readers process and reinforce their own worldview. And then there's the politics of it all, at least for this conversation. You see, one of the most sobering parts of this episode is Dylan's point that Trump's recent cuts to US foreign aid may well result in millions and millions of unnecessary deaths. It is one of those policy moves that might not be on the front page every day, but could quietly end up being one of the most harmful acts of his presidency. So, yeah, this episode ranges wildly from journalistic norms to artificial intelligence to even why Dylan defends The Last Jedi. But through it all, is someone who approaches politics and the craft of writing about politics with an unusual level of precision and care. So, yeah, you might be here for the deep dives into policy, or you might just be here because you wanna hear a thoughtful conversation about the world that we live in.
Andrew Xu [00:04:31]:
But regardless, I do think that you're going to enjoy this episode. Dylan brings a level of insight and clarity to these topics that is very hard to find, and I'm very grateful to have had the chance to speak with him. Hope you enjoy. Dylan Matthews, thank you so much for taking the time to come on to this podcast.
Dylan Matthews [00:05:08]:
Happy to do it.
Andrew Xu [00:05:09]:
So I'm going to begin this conversation by throwing you under the bus at least for some some small at least for some small cohort of listeners. Can you explain to the audience why you love The Last Jedi so much? And
Dylan Matthews [00:05:24]:
it's been a while since I rewatched it. You and I talked about this. There are themes in it that I liked a lot. I liked, sort of the subplot on the gambling planet and the sort of casual corruption, even though technically the empire has been overthrown and, you're supposed to be in a new regime with the Republic that sort of the ordinary problems of life in the galaxy go on. And I think that's sort of an interesting theme to explore. I liked the reveal that Rey was not a Skywalker at all, and there was nothing in particular special about her. And I think that, compared to things like introducing the midi-chlorians in the prequels, said interesting things about the ways her power and destiny work in the Star Wars universe. Yeah, I also just like Rian Johnson as a director.
Dylan Matthews [00:06:16]:
I like Poker Face a lot. Brick, I think is one of my favorite noir movies. And so I like the style that he brought to it.
Andrew Xu [00:06:26]:
Yeah. So I would love to argue with you more, but we have a limited amount of time here. So I guess yes. I'll I'll move on unless I can find a way back. Do you have superpowers that allow you to see to the future?
Dylan Matthews [00:06:39]:
I do not have superpowers that allow me to see into the future. Alas. Alack.
Andrew Xu [00:06:44]:
Yes. If you could have superpowers, what would they be?
Dylan Matthews [00:06:50]:
Oh, man. I think it's. Yeah. You have to sort of scope, scope this to some degree that, like within the universe of things that superheroes have, there's some things that clearly seem more valuable than others, like Ant-Man's skills seem less useful than Superman skills, Being able to sort of input arbitrary energy into any task at any moment seems pretty high impact. I don't know that it's absolutely the best. I always thought Magneto was kind of overpowered within the X-Men universe, and so I would go for something like that. Like, the ability to manipulate objects at will, telepathically, even objects within a specific category like metal, that seems, that seems pretty cool.
Andrew Xu [00:07:38]:
Alright. So I guess now I'm gonna try, in the same way that you tried scoping out superpowers, I'm gonna try to scope out your commentary a little bit. Sure. Would you say that the scope of your commentary has changed since the start of the year, or are you still on that policy wonk be talking about economics and AI and stuff?
Dylan Matthews [00:07:57]:
I'm doing a little bit of everything. I think, it's funny that you say AI is a policy wonk beat. I think of it as still a tech beat that is probably becoming more of a an economics and policy beat as it becomes a more normal part of everyday life. I'm trying to keep an eye on AI. I've written about budget and tax policy for years, and so, the current moment with the reconciliation debate seems really important and a moment to be thinking and writing about that. But I don't know. I've spent my whole career not really having a beat, and I don't expect to start now.
Andrew Xu [00:08:32]:
Yeah. So you don't have a beat. You don't stick to one beat. How about this? So I remember a while back, I spoke to Jessica Grose of the New York Times where she doesn't really have a beat, as well. She talks about tons of topics, whether it be feminism or education or religion or whatever. But she tries to have some kind of through line that connects them together about it's like about "the way that we live now" and how so many different facets of our culture are affecting the way that we live now and are affecting the way that we have community. So even if she doesn't have a beat, she does have some kind of throughline that she can very cheesily say to people. Do you have some kind of throughline that you can cheesily say to people about what connects all of your commentary?
Dylan Matthews [00:09:15]:
I've tried to again, I've tried to develop one for twenty years, and I don't quite have it. I usually say that I largely write about economics and social sciences, which is broad enough. And I think that it sounds specific to people, but if you know economics, you know that they believe that they can study absolutely everything and so can contain the entire universe in that. But that's usually what I say at parties. I think the tagline of Future Perfect, the section I write for is that it's about ways to make the world a better place. And so I will sometimes say that in more formal settings feels a little pretentious for person to person interactions. But, I think that's also a theme or a lens that I sometimes summarize my beat as.
Andrew Xu [00:10:00]:
Yeah. One thing I noticed about your writing is that you try to be as nuanced and precise as possible. Like, you try to possible. Like, you try to cover all the different corners that you can. Like, even if another writer would avoid writing caveats all the time because it sacrifices moral clarity, You always try to do that because you're trying to make the world a better place, and you prize accuracy. How do you how do you make your writing nuanced while also remaining engaging and convincing at the same time?
Dylan Matthews [00:10:28]:
This is a very flattering premise for a question. I'm sure I've fallen down on making it precise and I'm sure I've definitely fallen down on making it engaging, at times. Yeah, I think like that wasn't always a belief that I had. I think I was more like morally accepting of people who felt like they had to sort of sand down the edges of the truth or like summarize things more bluntly when I was younger than I am now. And, and I think some of that is just sort of seeing, witnessing the dynamic of, it's hard to tell to predict what lies are actually going to be good for the world to believe. And I think that's partly because it's very rarely a good thing for the world to believe a lie. And most people are not smart or perceptive enough to predict what's going to be beneficial in terms of sort of, being less than perfectly accurate.
Andrew Xu [00:11:32]:
Yeah. Straussianism was a bad idea.
Dylan Matthews [00:11:34]:
Yeah. I'm certainly not smart enough to do that. I think the other thing is that the people you, you fool are not people you need to persuade. They're people who are your allies anyway. So I think I think about here is the, the report that periodically comes out saying that most climate change is due to a hundred companies. And you dig into that and that turns out to be wildly misleading because most of the companies are oil companies, and they contribute because they sell oil to individuals and businesses who burn the oil to benefit themselves. So ExxonMobil: it's attributing to ExxonMobil the fact that I have occasionally refilled my tank at an ExxonMobil station. But it's not them.
Dylan Matthews [00:12:21]:
It's I have a demand for oil and as part of society. Anyway, no person who didn't wanna take action on climate change has ever been persuaded by that. It it a mind virus among people who already care, but find this helpful for connecting it to sort of economic priorities they like for thinking less about the individual impact of their own decisions and, giving them a reason to divert to thinking more systematically. So yeah, I think I've come to see patterns like that and think that they're they're broadly harmful. In terms of how do I do it? I think it's hard to write well unless you are a harsh critic of yourself. There's a Nate Suarez has a nice series called Replacing Guilt, on how to make guilt no longer a major motivation for for yourself because it it can be toxic in a lot of other ways. But I think, historically, that has been my tool is that I feel really bad if I, write something and I feel like it was it was glib or inaccurate or something, and it kind of haunts me.
Dylan Matthews [00:13:33]:
So I don't know that that's advice. I feel like as as Nate writes in those pieces, there are better ways to motivate yourself than guilt. But, guilt? Guilt is a powerful motivator.
Andrew Xu [00:13:42]:
Yeah. Well, I guess one tip that if Julia Galef were still here, maybe one thing that she would say that would be useful is that a lot of those people that are willing to improve over time is that they're willing to accept negative feedback because they feel proud of themselves about being willing to acknowledge their shortcomings. So maybe you could feel proud of the guilt that you felt, if that makes any sense.
Dylan Matthews [00:14:02]:
Sure. Yes. No. I being willing to acknowledge my own shortcomings has never been a one of my great weaknesses. I wish I had fewer shortcomings to acknowledge, but don't we all.
Andrew Xu [00:14:13]:
Yeah. But okay. So how about: there's one strand of one something that you said that I wanna pick up on about how maybe another younger, more boisterous journalist, they might avoid bringing caveats because they wanna improve moral clarity. Maybe it's easier to write that way. Maybe so many of us are so negativity bias pilled that it is hard for us to acknowledge the good things in the world. So, a), could you explain the what you would consider to be the prominence of negativity bias? And, b), could you help explain how that affects journalism in general?
Dylan Matthews [00:14:46]:
Yeah. So I think I wrote a piece about this a while ago. There seems to be a pretty persistent finding in social psychology, and I'll insert the caveat here that a lot of persistent findings in social psychology later turn out to not be true. So you should be skeptical of this one the same way you're skeptical of all of them. But I think so far it has replicated pretty well and has not fallen the way some other ideas in psychology have. Have. But a pretty persistent finding that people feel negative emotions, respond to negative message, messages sort of feel negatives more powerfully than they feel positives. And I think like on reflection, that's a phenomenon that I think most people can relate to that I am, I feel the quantity of happiness I get from someone praising an article is smaller than the quantity of unhappiness I get from someone criticizing it.
Dylan Matthews [00:15:40]:
The happiness I get when someone thanks me for a nice dinner is worth less than if someone says that dinner was terrible. And, so yeah. But bad is stronger than good is the way that one review article on this summarize it. And I think this affects people's willingness to to click on news. I think this is something I've experienced in my own writing. I think there has been good research. A guy named Stuart Soroka has done some interesting studies on this that just articles that are negatively framed do better, and attract more readers, connect to people more. And, yeah, I think that does, sort of, profoundly shape the way people think about the world.
Dylan Matthews [00:16:29]:
I think it also is interesting in that, I think, more educated, well read people tend to think of themselves as, able to avoid certain biases or ignorances. I think this is a problem that is, in some ways, even more pronounced the more educated, the more, the more you read the news, the more plugged in you are. In part because I think negativity or cynicism reads as sort of smart and sophisticated.
Andrew Xu [00:17:02]:
Like there's a default assumption that if you're more cynical, that means that you're smarter and that you're more nuanced or whatever.
Dylan Matthews [00:17:10]:
Right. You understand something beneath the surface. And the person who is really up on everything and thinks things are going well, that's a person who has not thought as as clearly and is refusing to look at the things that you've uncovered. And I think that bias is really profound that, this is a, a small example. But I've read four articles in recent months that casually say that Bill Clinton, decimated the safety net. And it's true that he cut one safety net program, aid to families with dependent children. He substantially expanded Medicaid, the earned income tax credit, created the child tax credit, created the children's health insurance program. On net, substantially expanded the safety net.
Dylan Matthews [00:18:07]:
And, in reputable publications, they will just say this. And I think it's like, the idea that things have just been going badly and there's a long term negative trajectory for poor people in America is just like a very hard narrative to dislodge mentally. And, in part because it seems like you're telling a sophisticated long term story about about the history of the country. Whereas if you say, you know, things have been gradually improving, they're not perfect, but they're improving, you sound sort of clueless.
Andrew Xu [00:18:44]:
Yeah. Okay. Then then I have an interesting question. So what do you say to people who are very casual observers of the news and they very occasionally see things on TikTok and maybe read the front page headlines? And they would say that so far, their guess is that the second Trump administration hasn't really been that bad. And the reason why newspapers say that it's so bad is because of negativity bias. What would you say to that?
Dylan Matthews [00:19:07]:
That's a fun judo move you're trying there. Yeah. I mean, I think there's definitely... I'm sure my assessment of things is inflected by negativity bias, the same as everybody else is. I think there's also there are things where Trump has done so much that you can probably pick out some cases and say like, isn't it overblown what liberals said attacking this? And I would probably agree with you. I get a lot of spam messages that are ostensibly from former labor secretary, Robert Reich condemning him for for trying to defund PBS and NPR. And I like PBS and NPR fine. It's not in the top 100 issues I would care about in the federal government. And so I would I would say that is as an example of negativity bias and fearmongering.
Dylan Matthews [00:20:03]:
I think there is serious people who know a lot about global health. The most recent issue of Asterisk had had some stuff on this, but serious people who know a lot about global health are predicting that millions, maybe tens of millions of people will die as a result of the PEPFAR cuts. Or PEPFAR cuts, cuts to aid, sort of general, foreign aid retrenchment under Trump. That didn't happen in his first term. I can't think of anything else that happened in his first term that had a toll like that. I don't think those estimates are a result of negativity bias. I think those are credible. Similarly, I mean, we'll we'll have to see where the tariffs land, but they didn't do tariffs of anything like the scale.
Dylan Matthews [00:20:45]:
They didn't cause anything like this much economic disruption. That seems another order of things. So I think there are areas where the outcomes seem clear enough and the differences seem clear enough that at some point, you have a ground truth. And it is just true that the tariffs are bigger than they were, and it is just true that they slashed foreign aid this time, and they didn't last time, really. So yeah. I think you it is not impossible to try to see the world and try to account for these biases. And I think when you do, there is a meaningful difference between this term and the previous
Andrew Xu [00:21:26]:
one. But do you think there's a meaningful difference in terms of how panicked journalists are covering this term compared to how panicked they covered the first term?
Dylan Matthews [00:21:35]:
Yeah. I think people are much less panicked this time. And like my memory of 2017 was that it was so people were so shocked that he won. People were there were so many leagues. There were so many sort of establishment Republicans in the administration who were vocally upset about things he was doing and leaking constantly. They had a special counsel appointed sort of within the first year. The degree of sort of panic and anger and fear was just much greater in 2017 than it is today. And I think that it's been particularly pronounced in thinking about the reconciliation bill is that is like, the house just passed legislation that is has very serious cuts to Medicaid and SNAP.
Dylan Matthews [00:22:31]:
And it looks like there is a margin for them to pass in the senate. We'll see. But them trying to repeal Obamacare in 2017 was the big story of the summer and constant coverage. It was top of mind, and they were literally doing this bill on the dead of the night. It seems like it was overwhelmed by things like Trump meeting with the president of South Africa or, the ongoing El Salvador immigration stuff. Just sort of people have not been focusing on that the same way. They haven't been panicking the same way even though I think the grounds for panic were as as strong or stronger on that specific issue.
Andrew Xu [00:23:15]:
Would you at least, so would you acknowledge that during Trump's first term, people were overreacting to Trump because of negativity bias? Would you consider that accurate?
Dylan Matthews [00:23:24]:
I think on some things like, there's a lot of people in the world. Like, I think there were definitely some things that I think the tax bill he passed in 2017 was bad. It was not like the apocalypse. I think that's an area where I think there was an overreaction. His dealings with Russia were genuinely bad and corrupt. And I it feels silly to go back and be like, people shouldn't have been mad that he was encouraging Vladimir Putin to hack his opponent's emails, and he was trying to interfere with investigations of him. And he fired the FBI director because he thought he was disloyal. Like, that stuff was genuinely bad.
Andrew Xu [00:24:05]:
Okay. So now if that was genuinely bad and now the concern is that millions and millions of people are gonna die, that does I think maybe we should lay some groundwork for that. So to begin laying the groundwork, could you give your somewhat hot take that George W Bush was not as bad of a president as many liberals think he was?
Dylan Matthews [00:24:25]:
Sure. So I think this is mostly an argument about PEPFAR. So, Bush, the big his biggest failure by far was was invading Iraq for no reason. I think the most credible estimates I know suggest that killed maybe 250, three hundred thousand people, maybe more. It's hard to get civilian numbers from that. It was completely unnecessary. Wasted lots of blood and treasure for no reason. Big, big, big failure.
Dylan Matthews [00:24:56]:
I think you have to put that, in weighing his legacy, alongside, PEPFAR, which was his plan to substantially increase US funding for anti HIV measures, in Sub Saharan Africa, which probably saved several million people's lives. And I would add to that he started the president's malaria initiative, which became a major funder of anti malarial bed nets and and, drugs. So I had I think it's fair to say that, George W Bush had the strongest foreign aid global health record of any president, at least since the Marshall Plan. And he did both these things. And so I don't think the takeaway isn't, PEPFAR wipes away the sins of the Iraq war. He would have have been a better president still if he had just not done the Iraq war, and it just did PEPFAR. But I think it puts him for me in a category more like LBJ where you have a set of tremendous accomplishments on on the one hand, the civil rights act, voting rights act, Medicare, and a horrific war on the other hand that escalated and that killed, vastly more people than Bush's war did. And I think the general historical memory of LBJ is not like, this guy is a horrible failure.
Dylan Matthews [00:26:21]:
It's "this was a complicated, man who was was great in some ways and horrific in others." And I think that's basically what what I think of George W Bush.
Andrew Xu [00:26:31]:
Okay. So let's say that George W Bush did save millions and millions of lives. What was the purpose of that? Like, what was the purpose of the US's foreign aid for things like PEPFAR?
Dylan Matthews [00:26:42]:
Do you mean what is the purpose in terms of US national interest? The purpose was to save millions of lives.
Andrew Xu [00:26:48]:
You know how cynicism like, people assume that cynicism is much smarter. Like, maybe someone would read that and they would immediately assume that a smarter take would be to be cynical and say that, oh, they were saving people's lives because it benefited their own self interest. Like, how much is that true, and how much do you think foreign aid was a product of actual altruism?
Dylan Matthews [00:27:12]:
I don't know of any serious security case for doing PEPFAR. I think maybe you could have come up with one in the Obama or early first Trump or Biden years based on China's investments in Africa and the need to sort of combat them and show that the US is still, sort of altruistic and interested in African issues. 2003, I think the general view was that, Sub Saharan Africa is not strategically important for The United States. It's not especially important to, retain the support of people there. This is not a major foreign policy security issue. The US did not get significant economic benefits from doing this. Most of the drugs are made in India, not in the US. I think the explanation that makes sense and that explains what happened, is the one that you get from most people, which is...
Dylan Matthews [00:28:10]:
George W Bush is a sincere evangelical Christian, who saw that millions of people were dying and thought it was his duty as as a Christian and a member of the brotherhood of man to help them. Yeah. I don't really know of a convincing explanation for why we did it other than that. And there were sort of secular people pushing for this who also care about global health for various reasons, but I think, yeah, it's very clear that George Bush's sense of morality is very religiously implanted. And I think that comes through when you talk about PEPFAR and when, when people who, worked in his administration try to tell that story.
Andrew Xu [00:28:48]:
How many Americans do you think were actually aware of the extent of Tthe US's foreign aid before Trump tore it down?
Dylan Matthews [00:28:56]:
I'd say very, very few. A famous finding in public opinion research is that if you ask people what share of the the US budget they think foreign aid is, they'll often say something like a quarter or a third rather than less than 1%, which is was the true answer. So I think there's a way in which people like, either people are vastly overstating this or people are bad at estimating those kinds of numbers and thinking about them. I don't think much of anybody had an accurate sense of
Andrew Xu [00:29:30]:
that. Okay. So there's one quote from one of your articles that I wanna quote right now because it talks about how the Trump administration was able to do these things on foreign aid. "The administration has seized control of spending from Congress, particularly on foreign aid matters. And so the bipartisan coalition that kept aid programs alive for decades has largely been helpless." So is the Trump administration successfully seizing control of spending from Congress? And if so, maybe the fact that so many people don't really know about these foreign aid programs or overestimated the extent of these foreign aid programs means that it was never politically popular and someone could have always done this.
Dylan Matthews [00:30:16]:
I think the degree to which they have tried to undo congressional control spending has ebbed since January and February at the height of the assault on USAID. So they've lost a bunch of court cases.
Andrew Xu [00:30:31]:
Could you could you explain USAID as well?
Dylan Matthews [00:30:34]:
USAID is the US agency for international development. It's our main foreign aid agency. So there are a lot of people in the Trump administration, notably Russ Vought, who runs the Office on Management and Budget, who believe that the president has, unilateral power to impound money appropriated by Congress, meaning that even if Congress says you have to spend a billion dollars on on this anti HIV program, the president has it within his power to just refuse to spend that. This is a crank legal theory that no one believes, and that I think has failed in the courts subsequent to them trying to do this. I think there are two additional things to say here. One is they've really tried it with USAID and they were doing doing things that were more clearly within their powers to get veteran staff to quit, to close down whole programs, to put things on pause. And I think they created facts on the ground that were hard to undo. And they didn't do that for other kinds of spending.
Dylan Matthews [00:31:47]:
They didn't go through the Agriculture Department and do that. They tried in the Social Security Administration, and ran into more political pushback just because, it's Social Security, and you had reductions in the ability to respond to seniors having problem troubles with the program. So they tried to do this with USAID. I think the reasons it succeeded there are a) foreign aid has never been especially popular, and there wasn't as much popular anger as there was over something like Social Security. b) I think the courts do say that the president has some authority and flexibility when it comes to spending on, foreign affairs. That that is closer to the inherent powers of the president. And foreign aid, you can you can make an argument is closer to the president in practice has the authority to order a military strike somewhere without going through the congressional appropriations process, and maybe this is more like that. And so I think there's a lot we still don't know.
Dylan Matthews [00:32:55]:
The courts operate very slowly, and they're still considering challenges to, the USAID cuts. I think with Elon sort of leaving DOGE and the DOGE process sort of looking like it's starting to wind down, It might not face as many, as many, future attempts as it saw in early this year. But I think it's clear that it was the area that they wanted to first test these ideas about impoundment. They hit it hard enough that even if courts rule against the administration going forward, it's gonna be hard to reconstitute USAID and other aid institutions as they existed before. Yeah. It was the first test ground, and I think had the most lasting damage as a result.
Andrew Xu [00:33:53]:
Yeah. And the remaining test grounds the fact that like, so for USAID, the damage has already been done. But for future test grounds, since they did it before, the courts can basically preemptively prevent them from striking them down. Is that what you're saying?
Dylan Matthews [00:34:08]:
Right. And and a lot of USAID operates through contracts and working with contractors. And like a lot of the contractors are just straight up out of business now. And so you can say the court can say the money the money is back. You can, you can, spend it now. But it's not possible to spend it if the contractor you were hiring to do it no longer exists.
Andrew Xu [00:34:35]:
That's interesting. Okay. So, in your article about USAID, you wrote something that I'm gonna quote right now. Quote, this is normally the section in the article where I tell you where US foreign assistance stands today, but the truth is that absolutely nobody knows for sure, end quote. So you wrote that back in March. Does anyone know for sure right now, or is it all in limbo and no one knows?
Dylan Matthews [00:34:57]:
I don't think anyone knows for sure. I think it is not it is not a priority of Congress in their spending process. Their, their current priority is passing this tax cut legislation and cutting Medicare and, Medicaid and SNAP. They haven't been focused on foreign aid. Foreign aid is largely discretionary, and so it happens outside of the budget reconciliation process. We have several months to go before the next, sort of continuing resolution fight over discretionary spending. I think that will add a lot of clarity. In the meantime, I think the Marco Rubio and the rest of the leadership at state that's been overseeing USAID and foreign a policy have been very clear that they're, they're just not going to allow, the spending that was going on prior to the Trump administration to keep going.
Dylan Matthews [00:35:54]:
So I think we can say confidently that we're in a new situation where dramatically less is being spent on foreign USAID, where, a lot of people are going to die unnecessarily as a result. I don't think we can say for sure what will be preserved, how it will be preserved, what the institutional arrangements at the end of all of this look like.
Andrew Xu [00:36:19]:
It's pretty sad that, yeah, millions and millions of people are gonna die from, the lack of US foreign aid. So on a on a completely unrelated note, could you Sure. Could you explain your views on artificial intelligence?
Dylan Matthews [00:36:33]:
My views on artificial intelligence? Do you care to narrow that down a little?
Andrew Xu [00:36:38]:
Okay. Do you think that a sufficiently regulated AI would be very good for the world?
Dylan Matthews [00:36:50]:
And and by AI here, you mean sort of something capable of doing anything that a human can do on a computer, anything a human can do physically? Can it be a software engineer?
Andrew Xu [00:37:02]:
Alright. Any anything a human can do on a computer.
Dylan Matthews [00:37:05]:
Okay. Yeah. I think I can imagine a situation where I can imagine a lot of situations in which, the net effect on the well-being of the human race is positive.
Andrew Xu [00:37:21]:
Do you I mean, can you is are those scenarios mostly the net effect is slightly positive or the net effect is orders of magnitude more positive than you could possibly think?
Dylan Matthews [00:37:31]:
That I could possibly think. I mean, I think there's are scenarios where it is largely positive. I don't think there are any scenarios where it is a Pareto improvement. I don't think there are any scenarios where everyone is better off. And I think most scenarios involve some people losing dramatically and there being a lot of social disruption. And, I think there are also a lot of scenarios where sort of the net effects are negative. But, but I have a hard time imagining a scenario where everyone the world over is so much richer that they're incredibly happy about it. I think we will probably have a situation where there are mass, whole sectors of the economy.
Dylan Matthews [00:38:14]:
They experience mass unemployment sort of like the China shock, but twenty, thirty times as large.
Andrew Xu [00:38:21]:
Okay. Can you imagine any scenarios where with sufficiently regulated AI, 90 to 95% of the human race becomes not just slightly better, but exponentially better. Can you imagine that scenario?
Dylan Matthews [00:38:33]:
Yeah. I think I can. There there is a lot of variability in the future on this, and so I can imagine a lot of scenarios. I think for that to happen, you have to have really dramatic levels of income redistribution, and/or, sort of redistributing equity in the firms that are the big winners in this, AI firms, chip manufacturers and things, such that a lot of their equity is socially owned. A lot of their income redounds to the general population. So that's not regulation, but it is a new fiscal regime in most countries, and it is the opposite of the direction that the US currently is going in. So I think you need that. I think you need pretty strong controls on misuse.
Dylan Matthews [00:39:28]:
I think the evidence that these will be very useful for doing cyberattacks, that they'll be superhuman at sort of hacking into various systems and getting people's personal information, that they'll be superhuman at designing biological weapons.
Dylan Matthews [00:39:45]:
The day we're having this conversation, two people were murdered outside the two people working at Israeli embassy in DC were murdered. There was someone who was shot, seeming to be attempting an attack at the CIA headquarters. A world where people like that can trivially create bioweapons that can kill millions of people does not seem out of the question. And so you would need regulations to stop that. There was a useful piece by a guy named Tom Davidson on coups in AI and, the way that sort of AI could help people who wanna seize control of governments or other institutions. You would need guardrails against that kind of misuse. You would need a lot of guardrails for this to go right. And I think one major cause of worry for me is that the trend in the US has been toward deregulation and toward a general sense that, if you let the sector continue the way it wants to, things will work out.
Dylan Matthews [00:40:44]:
And I don't think that's remotely true.
Andrew Xu [00:40:47]:
Maybe part of the emphasis towards deregulation is because of a fear of backlash. Like, you know how the Biden administration tried to regulate the tech sector at least a little bit, and then the response from a lot of the tech sector was to become radicalized into supporting Trump and giving them more money for the 2024 Election. Maybe that's also a concern.
Dylan Matthews [00:41:06]:
Yeah. I mean, I think the, I mean, the current trends are driven by Republicans because Republicans control the US government. The dynamic within the Democratic party is how much do we have to accommodate these tech figures before they go full fascist and try to end American democracy. And it turned out that what Lina Khan wanted to do was too far on that scale. And that if you try to do what she did, they will decide they don't believe in American democracy anymore and go through a massive right-wing turn. AI specifically, I think there are dueling impulses within Republican policy circles on this. I think, someone like Leopold Aschenbrenner, who I don't think is a Republican, but I think is broadly right of center, is very worried about China, has a kind of Manhattan project style vision for how AI should be developed, where it's a largely military US government project, undertaken in a race with the Chinese to try to defeat them. And that calls for a lot of regulations, and really intensive government oversight.
Dylan Matthews [00:42:24]:
And that's a very different vision from the Marc Andreessen, "let a thousand flowers bloom," let Nvidia send their chips wherever they wanna go, vision of things. And I think so far, the latter vision has been winning in the Trump administration. And I think in speeches from people like JD Vance, they clearly do not trust AI safety people. Like, maybe I'm being a little unfair, but I think that is the sense I've gotten from speeches of his on AI. But yeah, I wouldn't characterize the dynamic the way that you did as, sort of, how much do we have to suck up to these people without a backlash? Like, they are the backlash. They benefit from the backlash. The question is, now that we have power, how do we use it here? And since I don't think there's anyone in the tech sector who thinks they will be less regulated under a Democratic administration.
Andrew Xu [00:43:21]:
Okay. What do you think is the the best-case scenario version of AI policy that the Trump administration could enact?
Dylan Matthews [00:43:30]:
I think my super optimistic scenario in the Trump administration on this is they implement harsh export controls, on leading ships to try to buy time and sort of limit proliferation of dangerous models outside the US. They, using Trump's sort of pension for unusual diplomatic openings and visiting North Korea and things like that, try to forge some kind of deal with Xi on, on controlling Frontier AI, which I think is something that maybe he could pull off and Biden definitely couldn't have in that kind of Nixon goes to China way. He has expressed support for the idea of a sovereign wealth fund, which I think a lot depends on the execution. And there's a way that the US could set up a sovereign wealth fund where it is just like a corrupt slash fund for the Trump family. But there is also a universe where it is maybe in addition to being a corrupt slash fund for the Trump family, a way to invest in equities and, get capital income that will accrue if there is a rapid takeoff in AI capabilities that drives returns on capital and the value of shares in companies like NVIDIA and Google, to new dramatic heights. So I think that's what it looks like. I think the odds of getting there are not high, but I think Trump is a very weird guy, and he can be pushed in a number of directions. And I think there are directions that are better, for the near term future of AI than other ones.
Andrew Xu [00:45:22]:
You know, we we we keep talking about how Trump and, for example, the tech right? They are the backlash. And it's always a matter of what do you do when you become get in power? So you have a question that's like, largely unrelated, but I guess there's some kind of tangential tether. But let's say that Democrats do finally regain power 2026. They regain the power in 2028. Do you think that they should convict Trump and his inner circle and just throw all of those people into prison?
Dylan Matthews [00:45:52]:
I mean, I think it depends what they do. The thing I am most outraged about, that the Trump administration did is cutting foreign aid. I think there's things that they did that were illegal in a civil way. Like, I think it is illegal for them to cut a lot of that spending without Congressional approval, but that's not a thing you go to jail for. It's maybe a thing you get impeached for. It's definitely something you get sued for, and maybe something the courts will rule against you on. But it's not something you get thrown in jail for, and I would not support prosecuting someone for doing that. And more generally, I do not support prosecuting people, for policy disagreements they have with me.
Dylan Matthews [00:46:38]:
Like, I think people who wanna cut foreign aid from the US are very, very deeply and profoundly wrong. But it is their right to believe that and fight for it in a democratic society. And so I don't think they should be thrown in prison. All that said, maybe he commits some crimes. I don't know. I'm generally of the view that, elite impunity is a very bad thing, that we should have prosecuted people for the torture regime under, under Bush, that not doing that set a bad example, that Merrick Garland should have been more aggressive than setting up a special counsel investigation, into January 6. So, yeah, I think if they do further crimes of that nature, absolutely, they should be prosecuted for it. But I think "is it a crime" and "is it a thing that I think is severely damaging" are are two very different questions.
Andrew Xu [00:47:43]:
Okay. So what about the question of, like so so you know how the Trump administration like, you remember that time they purged the FBI of people that worked on January 6 cases? And right now, they're trying to fill the executive their branch with a bunch of MAGA loyalists. What do you think Democrats should do with those people when they get empowered? Do you think that they should do some kind of revenge purge where they purge people that refuse to acknowledge the 2020 election results? Or do you think that would just result in Democratic backsliding?
Dylan Matthews [00:48:13]:
This is the short answer is just I don't know. There are people who know a lot more about how to uncrop the civil service than I do. And this is a situation where I would ask Don Moynihan or or, another student of federal bureaucracy what the best, the best strategy is, I will say I remember De-Ba'athification, which you were probably too young to remember. But but in the very early days of the Iraq war, they tried to purge everyone who had been a member of the Ba'ath party, in Saddam Hussein's regime. They dissolved the army. They just got rid of a lot of institutions, and the result was less that they had gotten rid of people who were gonna sabotage things and more that there were just wasn't an Iraqi state that was capable of doing basic state functions. And so I would be worried about something like that, especially at a point where, the bureaucracy has already been weakened, pretty dramatically. But, yeah, this is an area where I would defer to people who know more than I do.
Andrew Xu [00:49:19]:
Okay. I have another complete pivot question. Do you think that human beings are naturally predisposed towards authoritarianism? And if so, do you think there's something that could be done about that?
Dylan Matthews [00:49:36]:
With the caveat that this is once again an area where you should talk to physical anthropologists, people who are experts in the near history of human evolution. My understanding is, I can't imagine what the selective pressure would have been that pushed humans into, trusting authoritarian leaders. Like, absolutely. And also because in the era when there was real selective pressure on humans, societies were so small that my intuition is that people are more resentful of tyranny the closer it gets to them. Like, if I'm living in a small town of 500 people and there's one guy who is making everyone's life hell in the town and is writing it dictatorially, you just overthrow him. It's a lot easier, and people have less tolerance for it. We don't really have a lot of experience being in polities that are millions of people large, either democratic or authoritarian. So I would be skeptical with the claim that people are predisposed to that.
Dylan Matthews [00:50:48]:
I would be skeptical of the plain claim that people are predisposed to democracy. I'm generally skeptical of claims that humans are predisposed to things that other advanced apes are not predisposed to or don't have a reason to be predisposed to.
Andrew Xu [00:51:04]:
Final question. So let's say that there are members of our audience that are genuinely horrified by the stuff that Donald Trump and his administration are doing right now and wish that there was some way that they could provide more support. What are some organizations that you would recommend that they donate to that can help counteract Trumpism in a productive way? Because I do think traditionally a lot of resistance pilled people have dwarfed Trump in terms of funding, but they have still failed anyway. What are some ways that they could donate money in that would be the most productive in counteracting Trumpism?
Dylan Matthews [00:51:39]:
Interesting. So I think in terms of the near term effects on global health, I would direct them to GiveWell and GiveWell top charities, like the Against Malaria Foundation, the Malaria Consortium, Helen Keller International and its Vitamin A supplementation. There are a lot of highly effective global health charities that can save lives, and, you will you personally cannot undo all the damage they did, but you can help. There are also, if you're interested in doing lobbying on that, there are there are groups that lobby for, more support for free global health. Groups like Results or the, Friends of the Global Fight, Global Leadership Council. There's a pretty robust network there, and I talked to those people, so I'm not gonna recommend that anyone give to one or another of them. But, but they're doing interesting work there.
Dylan Matthews [00:52:39]:
I would say that if you the way that you get power in the United States is when elections, and the place where a normal individual, can probably have the most, impact is not at, Presidential or Senatorial elections, but in something, things like state legislature races, things like, mayoral contests, things that are lower turnout, where less money is spent, where your marginal dollar can have more more of an effect. So I think The States Project is a group that does a lot of work trying to help Democrats win local elections. I, as a journalist, of course, can would never tell you what local party to vote for. But, if you are a Democrat who wants to win state legislative elections there, they might be an interesting organization for you to look into.
Andrew Xu [00:53:39]:
Dylan Matthews, thank you so much for taking the time to come back onto this podcast.
Dylan Matthews [00:53:44]:
Always a joy, Andrew.
Andrew Xu [00:53:49]:
Thank you for listening. If you enjoyed this episode and would like to hear more, make sure to follow this podcast on Spotify, Apple Podcasts, or anywhere else that you get your podcast from. Until next time. Goodbye.
Share this post