The Sentinel Podcast

Rai Sur
  • 0.0 (0)
  • EDUCATION

World-class forecasters make sense of the world and how events could escalate into globally-catastrophic events. blog.sentinel-team.org

Episodes

  1. 10/09/2025

    How AI Will Transform Military Command and Control - Paul Scharre

    A conversation with Paul Scharre, author of Four Battlegrounds: Power in the Age of Artificial Intelligence, who joins us to talk about * how AI’s superhuman command and control abilities will change the battlefield * why offense/defense balance isn’t a well-defined concept * “race to the bottom” dynamics for autonomous weapons * how a US/taiwan conlict in the age of drones might play out * and more… Links mentioned: Gradual Disempowerment, Swarms over the Strait Transcript Rai Sur 00:00:36 Today we’re speaking with Paul Scharre. Paul is the Executive Vice President at the Center for a New American Security and the author of Four Battlegrounds: Power in the Age of Artificial Intelligence. He also formerly worked at the Pentagon on emerging technologies. Welcome, Paul. Paul Scharre 00:00:52 Thanks for having me. Rai Sur 00:00:53 We’re also joined by Sentinel forecaster and co-founder Nuno Sempere, Sentinel Superforecaster Lisa, and Superforecaster Scott Eastman, whose specialty is in geopolitics, epidemiology, and AI. Welcome, Scott. Paul, what is command and control, and how do you see AI significantly altering it? Paul Scharre 00:01:18 The term command and control is used by militaries to describe the internal organizational and informational processes used to organize military forces and coordinate their behavior. Militaries work in a very hierarchical format, with teams, squads, platoons, companies, and battalions. The way they communicate has evolved from signal flags on the battlefield to radio communications, to today’s use of computers. Improvements in command and control can yield dramatic improvements on the battlefield, even if the military forces themselves are the same. We often think about the physical hardware of drones, missiles, or robots, and that’s absolutely valuable. But there are also potentially transformative effects in command and control. If a military can gather more information about the battlefield, make sense of it faster than the adversary, make better decisions, and execute those decisions in a coordinated fashion, it can have dramatic effects. This is particularly true if they can change the battlespace faster than the adversary can react, leaving the adversary constantly trying to respond to what happened six hours ago. Nuño Sempere 00:03:04 What’s a concrete example of this? Paul Scharre 00:03:11 During the US advance into Iraq in 2003, US military forces were advancing very rapidly toward Baghdad. Through a precision bombing campaign, the US disrupted the command and control of Saddam Hussein’s forces by taking out headquarters and radio communications. This created a situation where Iraqi commanders were reacting to events after they had already happened. They would identify the location of US troops, but by the time they reacted, the US forces had already advanced. That’s a potential advantage of having better situational awareness and command and control. Artificial intelligence has a lot of potential value in this space. Rai Sur 00:04:11 In Four Battlegrounds, you wrote about how AI might change decision-making and strategic planning at higher levels. What are the potential impacts there, and how could it change the look of warfare? Paul Scharre 00:04:25 Let’s imagine what this evolution might look like 15 to 30 years down the road as militaries integrate more AI, autonomy, drones, and robotics. Military advantage can come from better technology—better drones, larger production capacity, more robots on the battlefield. We tend to think about the physical object, which is important. But major advantages can also come from improvements in command and control, and we can see examples of this in gaming. Look at how AlphaZero plays chess. The pieces are the same for both the human and the AI, so there’s no material advantage. In many cases, the situational awareness is the same—they’re looking at the same board. Yet we’ve seen that AI is able to process that information better and more holistically than people. In games like StarCraft or Dota 2, the AI can see the big picture and comprehend it all. Across a number of games, AI agents can engage in coordinated, multi-axis attacks and balance their resources more effectively than people. This isn’t just in real-time strategy games; chess grandmasters have noted that AlphaZero can conduct attacks across the whole board more effectively. We’ve also seen AI systems engage in superhuman levels of calibrated risk-taking. In chess, this can look like ferocious attacks. In StarCraft and Dota 2, human players have talked about feeling constantly pressured by the AI, never having a moment to rest. In poker, AI systems bet differently than humans, engaging in wild betting swings that are hard for even cold-blooded poker players to manage because of human emotions. There’s an information processing element, but also a psychological one. In combat, you get lulls in the action because people need to rest and reset. AI systems don’t. In the command and control aspect, AI has the potential to not just be better than humans, but to transform the very strategy and psychology of war. Rai Sur 00:07:53 What does this look like in practice, at the level of grand strategy, where generals use game theory and deception? What does adoption look like there, and what is the minimum capability required? Naively, it sounds like an AGI-complete problem. Paul Scharre 00:08:27 There are many beneficial things militaries could do on the command and control front that are not AGI-complete problems. If you hand over all your military forces to an AI, you need a lot of trust in its intelligence to handle that complexity. But you can imagine very narrow algorithms that optimize logistics or communication on the battlefield, which would have significant advantages. You could also have more advanced algorithms that assist with planning military operations while commanders remain in the loop. The AI could generate plans for offensive or defensive operations, and commanders would still review and decide what to execute. Over time, they might see that the AI’s plans are better than what humans come up with and decide to trust them more. Rai Sur 00:09:29 Who is best positioned to gain these advantages sooner rather than later? Paul Scharre 00:09:35 In general, more technologically advanced militaries have some advantages, but we’ve seen AI technologies diffusing very rapidly. The biggest limiting factor for militaries isn’t getting access to AI technology, but finding the best ways to use it and transform how they operate. The history of military-technical revolutions shows that what matters is not getting the technology first or even having the best technology, but finding the best ways of using it. An interesting historical example is British innovation in carrier aviation during the interwar period. The British military was ahead in inventing carrier aviation but fell behind Japan and the United States due to bureaucratic and cultural squabbles. It was primarily a technology adoption problem, not a technology creation problem. This suggests that militaries in active conflicts, like we’re seeing in Gaza and Ukraine, are much more incentivized to overcome these barriers to adoption. They’re also getting real-world feedback on how effective these technologies are, whereas getting those feedback loops in peacetime is much more challenging. Nuño Sempere 00:11:22 So you would greatly discount which countries have top AI labs as a factor, because you think adoption feedback loops are more important? Paul Scharre 00:11:32 Yes, absolutely. The top US AI labs are maybe a few months ahead of the top Chinese labs—perhaps 12 months, but not three years. The US military is, charitably, five years behind leading AI labs in adopting the technology, and realistically more like 10 years behind in actually using it. It’s not enough to have an image classifier on your drone video feed; you also have to use it to transform your intelligence operations and make your analysts more effective. That’s a much longer process, particularly in peacetime without competitive pressures. Having leading AI labs in your country is probably not the most important factor for battlefield operations. For example, what Ukraine is doing with drones is far more innovative than what the US military is doing. That pressure is more important. The exception might be in niche areas like offensive cyber operations. There, you might imagine that access to the most advanced AI systems could provide large, step-function increases in capability. If the NSA has a 12-month head start on Chinese competitors and can actually employ that advantage, that’s a place where a 12-month lead could matter a lot. Rai Sur 00:13:21 And if the difference-maker is having a place to apply it and get feedback quickly, cyber operations don’t require a hot conflict. You are constantly searching for intelligence and testing capabilities against adversaries, so those feedback loops are always active. Paul Scharre 00:13:44 You have more feedback in cyberspace during peacetime because of the constant engagement between intelligence services. That engagement may not be as robust as it would be in wartime—many constraints would come off in a wartime environment, leading to more cyber operations and tighter feedback loops. But you probably have more feedback in cyber operations now than you do in ground warfare for the US military. Scott Eastman 00:14:26 My understanding is that tens of thousands of drones might be used in the Ukraine war in a 24-hour period. Is a primary purpose of AI to assimilate all of that information in a meaningful way? An individual soldier might have access to their own drone feed, but how do you integrate that across the entire battlefield? Is AI critical for filtering that much information into something usable? Paul Scharre 00:15:02 This is probably

    1h 30m

  2. 09/11/2025

    Dangers of Recommender Systems - Ivan Vendrov

    Please subscribe to our new YouTube channel here!We speak with Ivan Vendrov about recommender systems and their impact on human attention and society. Five years after his post on aligning recommender systems, as they could be significantly augmented by generative AI, we revisit how these algorithms shape billions of hours of human time daily. Topics The Current State of Recommender Systems * What recommender systems are and how they function as "prosthetics for attention" * The unclear evidence on political polarization and echo chambers * The rise of short-form video and increased addictiveness * Saturation effects in human attention markets Incentive Structures and Business Metrics * How tech companies actually make decisions about algorithm changes * The tension between engagement metrics and user wellbeing * Why companies aren't incentivized to increase users' earning capacity * The multipolar trap preventing better alignment Future Trajectories and AI-Generated Content * The shift from algorithmic curation to algorithmic generation * Hyper-personalized AI companions and relationships * Long-horizon reinforcement learning in recommendation algorithms * The potential for recommender systems to shape human preferences Cognitive and Social Impacts * "iPad babies" and developmental concerns * LLM-driven cognitive atrophy and psychosis cases * The loss of shared cultural canon and common knowledge * Labor force participation and the unemployed/elderly as target markets Solutions and Opportunities * The massive waste of human potential in current systems * Ideas for aligned recommender systems using modern AI * Subscription vs. advertising models for better incentives * The need for diverse, community-specific platforms Broader Implications * Civilizational feedback loops and international competition * The vulnerability of human minds to hypnotic patterns * Network effects preventing innovation * The grief of losing aspects of human experience to AI Transcript AI-generated. Will differ slightly from the real conversation. Rai Sur 00:00:47 Today, we're speaking with Ivan Vendrov. He's a former AI researcher at Google and Anthropic and currently works at Midjourney, but we mainly know him from a post he wrote in 2019 on aligning recommender systems as a cause area. A lot has changed in AI since then, which has implications for recommender systems, so we wanted to revisit this topic with him. Welcome, Ivan. Ivan Vendrov 00:01:13 Thanks, Rai. It's good to be here. Rai Sur 00:01:14 We're also joined by Sentinel co-founder and forecaster Nuno Sempere, and Sentinel forecaster Vidur Kapur. Vidur Kapur 00:01:22 Hi, it's nice to be here. Rai Sur 00:01:23 Ivan, to start, what is a recommender system, and why should we care about how they behave? Ivan Vendrov 00:01:29 I like to think of a recommender system as a prosthetic for your attention. The internet is vast, and you need machine help to figure out what to pay attention to. Take the YouTube recommender system, which is probably the most important and most used one. YouTube has billions of videos it could show you, and no human could possibly watch them all to decide which are best for you. That's why we need machine help. A recommender system takes millions or billions of possible items—in this case, videos—and filters them down to a small set of one to five that we expect the user will like. This is typically done by training a machine learning model to predict observable user behaviors. The model predicts the probability that if we showed you a certain video, you would click on it, how long you would watch it for, and even whether you would click through any ads. As for why this is important, billions of hours of human time are allocated by recommender systems every day. It's staggering. Rai Sur 00:02:54 When you wrote that post, what issues did you see? Which ones have come to pass, and what new concerns have emerged for the future of recommender systems? Ivan Vendrov 00:03:06 I wrote that post in 2019 amid a lot of concern about addiction and political polarization—how recommender systems might be creating echo chambers and amplifying radical extremist positions. The sad thing is, we still don't know if that's true. We just don't know what impact recommender systems have had on our political system. People say we're in a more polarized phase where it's harder for us to hear each other, but I have not seen compelling quantitative evidence of this. In fact, there's some evidence that YouTube shows people more opposing viewpoints than they would otherwise get. Usually, people subscribe to one source, like Breitbart or *The New Yorker*, and only get one set of political opinions. But YouTube will sometimes show you something entertaining from the other side. So the evidence is still mixed. A big part of the problem is that we don't have the data. It's trapped inside big tech companies that have strong incentives not to analyze the impact of their algorithms, which could lead to legal liability. So we're in this weird epistemic position. These systems are incredibly powerful and allocate huge amounts of human attention. It would be amazing if they weren't causing all sorts of problems—and benefits—given their influence. But we just don't have a great map of what they're actually doing. Nuño Sempere 00:04:58 What about addictiveness? Have these algorithms become much more powerful since 2019? Ivan Vendrov 00:05:06 Again, we don't have great data on this. Anecdotally, it seems like short-form video represents the biggest increase in addictiveness. You can see this if you give a child YouTube Shorts—which I don't recommend doing—they will be hypnotized by the screen for a very long time. I've personally experienced falling into a TikTok or YouTube Shorts rabbit hole. It feels like a different level of addictiveness. Three hours later, I'll snap out of it and think, "What happened to my brain? I wasn't even enjoying that." So there's qualitative evidence. One of my concerns is that people go through phases where they simply need a distraction. If they weren't watching YouTube Shorts, maybe they would be doing something worse, like watching TV in a zombified state for seven hours a day. So it's not clear what the counterfactual is or what the net increase in addiction really is. But we have certainly gotten very good at creating addictive digital experiences. Rai Sur 00:06:30 I hadn't appreciated how unclear the impact is until you brought up the counterfactual of what people might otherwise be doing with their time. I have strong intuitive aversions to what many recommender systems are doing, and I act to curtail them in my own life based on those intuitions. But it's possible we're scapegoating them too much. It's hard to remain open about it. Nuño Sempere 00:07:06 On our side, we've also noticed anecdotal evidence. People seem much more glued to their phones. I hear anecdotes about people spending hours each day on YouTube or becoming more addicted to Twitter. Again, this isn't scientific, just anecdotal warning signs. For instance, we know one of the biggest YouTube channels is Cocomelon, which is known for hypnotizing babies. Rai Sur 00:07:47 My interest in this was renewed by a tweet from Andrej Karpathy. He mentioned that now that video and audio generation are unlocked, we can optimize directly for them. To date, short-form video platforms have relied on human content generation followed by algorithmic curation. Now, we can have algorithmic generation and curation, optimizing the entire process from end to end. What do you think about this development and the impact it might have? Ivan Vendrov 00:08:23 The best analogy is the difference between the internet and language models. For a long time, you could find a community online that produces human-written text tailored to your interests. But with language models, we can get even more hyper-personalized. A language model can understand what you're going through, use specific information about you, and give you the exact sentence you need to feel good in that moment. This leads to worries about LLM companions replacing real relationships or even LLM psychosis, where AI encourages people to go into strange rabbit holes by amplifying their existing beliefs. It seems a really addictive consumer experience will become possible through a hyper-personalized feed of content. Going back to basic human needs, this probably won't look like a TikTok feed with a bunch of different people talking to a screen. It will more likely be a relationship. People will form connections with one or a few AI personalities because what people find compelling is building a relationship and feeling deeply understood. I imagine that at different points of the day, depending on your emotional needs, you might be shown a wild, AI-generated feed that keeps you hypnotized. At other times, you might interact with an AI companion or girlfriend, developing a close relationship. Perhaps you'll be on video with them so they can read your micro-expressions and you can read theirs, creating an intense, full-bandwidth experience. Rai Sur 00:10:17 Let's discuss the qualitative considerations for forecasting the trajectory and impact of recommender systems as they become more powerful. A simple story would be to look back 10 years, see that people are spending more time on them, and extrapolate that line until everyone becomes a zombie. But that's too simplistic. There are other factors at play, like feedback loops, discontinuities, and incentives. What are the important considerations or dynamics we need to track to understand how this will play out? Vidur Kapur 00:10:59 One consideration is whether there will be any large-scale backlash. For example, Australia is trying to ban social media for people under 16. Will that approach spread elsewhere? Even without top-down regulation, there's the question of a potential culture shift against

    1h 9m

  3. 06/27/2025

    Iranian Regime Change? - Unpacking Broad Disagreement Between the Forecasters

    Transcript The following transcript is AI-generated and may differ from the original in slight ways. Welcome to the Sentinel podcast, where top forecasters discuss the potential for global catastrophic risks. I'm your host, Rai, and I'm joined today by my co-founder and Sentinel forecaster, Nuno Sempere, Sentinel forecaster Vidur Kapur... Vidur 00:00:44 Hello. Rai 00:00:45 And Sentinel forecaster Lisa. Lisa 00:00:47 Hello. Rai 00:00:48 Today we're going to be talking about the possibility of regime change in Iran. Before we go into that, does anyone want to describe why we care about this possibility? Why does it matter for global catastrophic risks? Vidur 00:01:00 From my perspective, regime change would reduce the probability of a nuclear detonation in the Middle East over the next decade. While it's possible it could lead to a more extreme, less rational regime, my base case is that it would reduce the risk of a nuclear exchange. That's why it has a bearing on global catastrophic risk. Nuño 00:01:27 In addition, Iran is a country of 90 to 92 million people, so regime change there affects a large population. Even if it doesn't lead to a nuclear weapon, we've seen that regime change can destabilize a region. With the change in Afghanistan, for example, terrorist groups in Pakistan could operate more freely. If the Iranian regime falls without a clear successor, the whole region could be destabilized in a way that leads to bad outcomes. Vidur 00:02:03 It can also destabilize other regions. The regime changes in Libya and Iraq, for instance, destabilized European politics. So the effects can be felt both within the region and elsewhere. Lisa 00:02:18 I'd like to add the flip side. Imagine an Iran that doesn't fund proxies throughout the region. While there would be immediate instability for a few years after regime change, as you've all mentioned, it's also possible that in the longer term, it could increase regional stability. This would mean a potential for less conflict, independent of the risk of nuclear detonations. Rai 00:03:05 Let's get to the forecasts. We're forecasting the probability of regime change on two timelines: by the end of August and by the end of the year. Let's go around and get your two probabilities, and then we'll discuss them. Let's start with Nuno. Nuño 00:03:32 For the end of the year, I'm in the 10% to 40% range, so I'll put it at 25%. I'm confused because Vidhur and Lisa, who are great forecasters, are at different extremes. To resolve that, I went back to basics and looked at the base rate of regime change. Regime change happens relatively frequently in countries around Iran. In the last 25 years, Afghanistan has switched hands, Syria's regime fell, you could argue Lebanon had a partial regime change, and Iraq was invaded. Based on that, I get a base rate of 2% to 4% per year, which is a good starting point. Of course, recent events increase that probability. But by how much? A 10x increase, from 1-in-50 to 1-in-2, seems like too much. That's how I'm thinking about it. Rai 00:05:15 And do you have a forecast for the end of August? Nuño 00:05:17 By the end of August is trickier. I'll go with 7%. I'm very uncertain about the timing of a potential collapse. More time allows instability to compound, but I also sense that the critical moment is now, rather than later in the year. Rai 00:05:43 What about you, Vidhur? Vidur 00:05:45 By the end of August, I'm at 10%, and by the end of the year, I'm at 15%. Rai 00:05:53 And Lisa? Lisa 00:05:56 I'm struggling a bit because so much is happening so quickly. It's difficult to figure out what's going on, especially since the stance on regime change from Israel and the US seems to shift daily. However, my forecasts are substantially higher than my colleagues'. For August, I would say about 50% to 60%. By the end of the year, 60% to 70%. I see much of that risk as front-loaded. The greatest period of instability is now, in the coming weeks and months, as the regime struggles to adapt after Israel has killed a large number of its leaders in airstrikes and drone attacks. The regime is very repressive, and it takes a lot of effort to maintain control. They seem to be trying to cope by increasing arrests. Another key aspect is Khamenei. It's not clear where he is, and he might be quite ill. I'm not sure he will be living very long, but we can talk more in a few minutes. Rai 00:07:45 You should interrogate each other about why you see the probability so differently, especially on such a short timeframe. Lisa 00:07:56 I'd like to say more about why I think the probability is so high. A big factor is that Khamenei isn't very present. I keep seeing reports—and I don't know if they're true—that the IRGC and other leaders are withholding information from him and that he's not fully involved in governing. For instance, after the recent ceasefire, he never came out to address the country and claim a glorious victory for Iran. It's just been crickets, which is weird. Nuño 00:08:41 How old is he right now? Lisa 00:08:42 86. Nuño 00:08:43 So that's what, a 5% chance of dying per year? Something in that ballpark. Lisa 00:08:48 I think there's something bigger going on here. Israel and the US do not want Iran to acquire nuclear weapons and have gone to great lengths to damage its nuclear program. And yet, regime leaders are doubling down, resuming enrichment, and denying IAEA inspectors access. There was even talk about leaving the Non-Proliferation Treaty. At the same time, Israel has stopped its strikes, and the US goal seemed to be simply dealing with buried nuclear facilities. It's hard for me to understand why they would go to all this trouble to set things back by only a few months or years if the regime is just going to keep pressing forward. One answer I can come up with is that they believe regime change is likely. Perhaps they have intelligence that Khamenei is on his deathbed. That's what I'm wondering. Of course, there are many factors, and nobody wants chaos. Rai 00:10:22 When you say setting things back by months to years, you're talking about the nuclear program and enriching uranium to weapons-grade levels, correct? Lisa 00:10:31 Yes, and generally, everything that goes into building a nuclear weapon. Rai 00:10:37 If the short setback—only a few months to a few years—makes you think this is about regime change, why doesn't it push you toward the conclusion that the strikes were simply ineffective? Why wouldn't that ineffectiveness, intentional or not, be the primary explanation for the short setback, rather than the focus being on regime change? Lisa 00:11:10 That's a great question. It's quite possible the strikes were the best they could do and that all they can hope for is to set things back by a few months or years. We could just be doing this on rinse and repeat. However, that strategy would become less effective over time. The more Iran anticipates such strikes, the more they would prepare by building deeper facilities and hiding them better. There's already talk of facilities that weren't hit. Iran could do many things in this cat-and-mouse game to make successive strikes less effective. That's why I would have thought they'd be more likely to push for regime change, or at least a change within the regime. Trump said regime change was on the table, and I believe the Israeli Defense Minister said that Khamenei could no longer exist. It looked like they were going for that. But it's also quite possible this was simply an attempt to address an immediate threat, and now that it's addressed, they're done. Still, we have to look at the regime itself. You can't go through these leadership changes without creating internal chaos. Their stability is weakened. The question then becomes what opposition groups, inside and outside Iran, do to take advantage of this situation and what help they might be getting from countries like Israel or the US. None of us really knows. Nuño 00:13:10 I would love to hear from Vidor on the opposing case. Vidur 00:13:13 To explain my reasoning, I also started with a base rate. Nunu's rate of 4% to 5% per year isn't far from my own. You can look at two categories of regime change: externally forced and internal. Since World War II, the base rate for an externally forced regime change in Iran is around 1% to 2%. This happened with the 1953 coup sponsored by the US and UK. The base rate for an internal change is also around 1% to 2%, as seen in the 1979 revolution. I'd give an external regime change a 2% chance in the next six months and an internal change a 12% to 13% chance. I'm low on an external regime change now because the US hasn't truly committed. You could interpret Trump's Truth Social post as saying the Iranian people might change their own regime to "make Iran great again," similar to how he views his own election. He later said he's not in favor of it. While the Israeli Defense Minister was aggressive, Netanyahu said the Iranian people might have to rise up themselves, implying Israel won't do it all for them. Neither Trump nor Netanyahu seemed fully committed. Yes, the strikes have destabilized the regime by killing top people, which could lead to an internal regime change. However, these senior people can be replaced. In a country of 90 million, there will be plenty of people willing to take those jobs, despite the elevated risk of death. We also haven't seen the kind of mass protests that would suggest a movement against the regime. We saw protests in 2009-2011, but nothing happened. The only recent protests I'm aware of were student protests against the US and Israeli attacks, which the Prime Minister joined. As Lisa pointed out, it's possible that the Ayatollah dies in the next six months. He is 86 years old, and an American male of that age has about a 5% to 6% chance of dying in that timeframe. You can argue he has better healthcare, but he has also had past health problems. His death

    52 min

  4. 05/31/2025

    Endgames and escalations of the US and Japan's expensive debt problem – Itay Vinik

    Itay Vinik is the cofounder and Chief Investment Officer of Equi, an alternatives investment fund that deeply considers rare market conditions in its investment strategies. He was was the cofounder of a long/short volatility hedge fund which made a famous bet on the 2018 volmageddon event. Transcript Rai 00:00:56 Welcome to the Sentinel podcast, where top forecasters discuss ongoing events with a view toward global catastrophic risks. I'm your host Ry, and I'm joined by Sentinel co-founder and forecaster Nuno Sempere, and Sentinel forecaster Lisa. Our guest today is Itay Vinik. Itay is the co-founder and Chief Investment Officer of Equi, an alternatives investment fund that deeply considers rare market conditions in its investment strategies. He was also the co-founder of a long-short volatility hedge fund, which made a famous bet on the 2018 Volmageddon event. Welcome, Itay. Itay 00:01:32 Great, thank you. Rai 00:01:33 Today, we will discuss recent indications of global economic uncertainty, primarily in sovereign debt markets. This is important because large economic struggles—like sovereign debt crises, austerity, and significant inflation in the world's largest economies—can cause shifts in political power and state capacity. These shifts can ultimately affect the trajectory of technology development, political doctrine, and war. As a recap, weak demand for sovereign bonds sent 30-year US Treasury yields briefly above 5%, German 30-year bond yields to roughly 3%, and Japanese 30-year bond yields to record highs. Many factors were at play: sticky inflation, significant government borrowing in recent years, a costly bill passed by the US House of Representatives, and Moody's recent downgrading of US debt. Persistent debt issues can put a lot of pressure on societies. What types of political shifts might we see in a world where financing this debt is not cheap? Itay, since Sentinel and Equi are both concerned with escalations into longer-tail outcomes, what are some of the more impactful economic issues we might see from here? Itay 00:02:43 The "bad outcomes" that can happen include a haircut default-type cycle. In this scenario, if governments let the market operate freely, higher real yields (yields above the inflation rate) would kill long-term investment. There would be less incentive to invest, and capital would flow back into Treasuries. This typically results in deflation, naturally resolving the debt through defaults. Yields would eventually come down, but with devastating economic consequences. The second choice, seemingly more popular with policymakers, is to inflate the debt away. By devaluing money, they effectively "kick the can down the road." These are the two main choices. In my opinion, given the option, policymakers will always choose to kick the can down the road. However, in extreme tail-event scenarios, policymakers might not have that choice. We can discuss that; it's an edge case, but possible. Lisa 00:03:54 Itay, could you ever see a scenario with an actual default or a haircut on US Treasuries? Personally, I can't envision that. I'd like to hear your opinion. Itay 00:04:06 I don't really foresee that. The US is quite different from the rest of the world because, at least for now, it has the global reserve currency. As long as the US can create the global reserve currency, the likelihood of a default on US Treasuries is minimal. However, in the distant future, if that were not the case, a default would certainly be possible. Lisa 00:04:26 Yet, we see the prices of credit default swaps on U.S. Treasuries going up, so apparently not everyone feels that way. You're saying there's very little risk as long as the dollar remains the world's reserve currency, but we're seeing the world start to pull back from the dollar. Personally, when I look out over the next decade, I see a very low risk. Over extremely long time horizons, I think that could change, obviously. But at least for the short to medium-term future, that doesn't look like a serious risk to me. Rai 00:05:06 Let's go around to the forecasters and get a preliminary forecast. It's tough to operationalize this for the US, in particular, because they could monetize the debt. But let's try for some fuzzy operationalization of what a US default might entail. Lisa 00:05:17 You mean a US default, a straight-up default of some sort? Rai 00:05:22 Yes, but does that fully capture our concern? They could technically avoid a default by monetizing the debt. However, if the US monetizes the debt, that's still a scenario we're worried about. Nuño 00:05:33 When you say monetizing the debt, what does that mean? Lisa 00:05:36 Inflating it away. Nuño 00:05:37 Okay. Lisa 00:05:37 To me, inflating it away is infinitely less dangerous than investors taking a haircut. That would destroy Treasuries as a safe-haven asset. Itay 00:05:50 Consider what we saw this past April: what I call a "triple yazoo" moment, a term coined after the Japanese crash, where the stock market, currency, and bond market crashed simultaneously. That is pretty unusual for the United States. There was also a moment concerning the basis trade, which is how hedge funds lever up with long-dated Treasuries. If you compare this to other crises, like 2008-2009 or 2011, it's different. 2011 is particularly interesting. In August 2011, S&P downgraded US credit for the first time. The stock market crashed 20%, while TLT, a good measure of long-term bonds, went up 15-20%. Contrast that with recent events after "Liberation Day": the US bond market sold off hard, along with stocks and the US dollar. This is an unusual scenario. Lisa 00:06:50 I completely agree that was extremely unusual. But we didn't come to the brink of an actual default. To me, an actual default means a "not getting your money back" type of scenario. I'm not talking about merely losing money in the bond market. Itay 00:07:08 I'm thinking about the safe-haven aspect and the typical investor mindset where assets are anti-correlated—when one goes up, the other goes down. These correlations broke. Typically, long-end bonds would do well because when stocks crash, the Fed is expected to cut rates and the economy to slow down, meaning bond yields should go lower. We haven't seen that in this recent move, which is concerning because it suggests the bond market is more fragile than usual. And I agree entirely: the risk of an actual US default, US Treasuries taking a haircut, or the US losing its reserve currency status is, at this point and for the foreseeable future, minuscule to non-existent. Nuño 00:08:04 I'm less familiar with the numbers, but I'd initially estimate 0.5% or 0.1% a year. Does that seem significantly off, or would you suggest less than 0.1%? Lisa 00:08:16 What actual default rate are you considering? Itay 00:08:19 It's hard to say, but I'd consider it an edge case, probably around 0.1%. The global financial system's infrastructure and plumbing are fully built on U.S. dollars. Approximately 65% to 70% of all FX reserves are still in U.S. dollars. Furthermore, 80% to 90% of transactions in commodities markets and other infrastructure are conducted in dollars. Saudi Arabia's recent deal with the United States reinforces the petrodollar system, ensuring dollars continue to be used for oil exchange. For example, if a Norwegian company digs for oil in the North Sea and borrows money, they most likely borrow and secure it in dollars. A vast amount of dollar-denominated debt is also held between countries and by emerging markets with U.S. banks. Because this entire system operates in U.S. dollars, replacing it is extremely difficult and will likely take many years. When people discuss the end of the U.S. dollar, the question becomes: what is the alternative? What could replace the vast existing dollar-based infrastructure? Currently, no pure substitute exists. It would be a very slow unwinding of the current system. This is a matter of decades or more, not something in the near term. Nuño 00:09:54 Regarding the long or medium term, how do you define those? Is it a couple of years? Five years? Are we looking at 0.1% for five years, or for 50 years? As a trader, what do you mean by long term and medium term? Itay 00:10:11 It's somewhat arbitrary, but the longer the time frame, the greater the uncertainty and the higher the probability of unlikely events. For the next five years, I'd estimate 0.1%, increasing thereafter. It probably wouldn't be significantly more for 10 years. Beyond 10 years enters the medium term. Twenty to thirty years from now, all bets are off; it's very difficult to predict. Global reserve currencies have historically lasted between 120 and 200 years. However, things are moving faster now. Before the U.S. dollar, the British pound was the reserve currency, used in transactions. The Netherlands and Spain also had periods where their currencies were dominant. These shifts occur based on the dominant global power of the era. For now, what business or company would agree to transact in, say, the Chinese yuan at scale? Perhaps Russia is compelled to due to sanctions, but widespread adoption by others is unlikely. It's hard to imagine companies widely adopting a substitute for the U.S. dollar. However, we are slowly transitioning from an American-dominated world to a more multipolar one, offering more alternatives and options. But I don't see an immediate systematic threat that could dismantle the current system. That said, high debt levels and bond yields do pose a substantial threat to the economy. Rai 00:11:45 We've discussed the U.S., Itay. What are your thoughts on the situation in Japan? What is the impact for Japanese investors and policymakers? What do you foresee? Itay 00:11:59 This is a very loaded and complicated topic. In historical context, Japan, the third or fourth largest economy in the world—it fluctuates a bit, and also depends on the exchange rate—is the first to experim

    46 min

  5. 05/09/2025

    The Risk of Nuclear War Between India and Pakistan

    Nuño Sempere, Vidur Kapur, and Lisa discuss the possibility of nuclear escalation between India and Pakistan. Timestamps (00:16:00) - Intro (00:43:00) - Background on the conflict (02:26:13) - Direct forecasts of nuclear weapons use (15:45:10) - The role of other nations (20:26:09) - What if Pakistan collapses? (22:45:16) - Russia's incentives in global conflicts (25:40:08) - What are the forecasters looking out for next? (29:28:06) - Outro Transcript The transcript is AI-generated and slightly differs from the phrasing used in the recording. Rai Sur 00:00:16 Welcome to the Sentinel podcast, where top forecasters discuss ongoing events with a view towards better understanding global catastrophic risk. Today, we'll be talking about the recent military escalation between two nuclear weapons states, India and Pakistan. I'm your host, Rai, and I'm joined by three of Sentinel's world-class forecasters: Nuño Sempere, who's also my co-founder at Sentinel; Lisa, also known by her forecasting moniker, Be Like Water; and Vidur Kapur. The situation is evolving rapidly, and we are recording this on Wednesday, May 7th. To ground the discussion, the focus will be on nuclear risk, as Sentinel is concerned with escalations in the long tail of extremely bad outcomes. Most of that risk here is nuclear. About two weeks ago, a terrorist attack in Jammu and Kashmir, a contested territory on the India-Pakistan border, killed 27 people (mostly Hindus) and injured 20 others. This caused an immediate diplomatic breakdown that has since escalated to military action at a scale not seen since the last Indo-Pakistani war in 1999. India's nuclear program conducted its first nuclear test in 1974, and Pakistan had theirs in 1998, but neither has tested a weapon since 1998. The countries have very different nuclear doctrines. India committed to a no-first-use doctrine in 1998, meaning they pledged not to use nuclear weapons proactively. Pakistan, however, affirmed as recently as this conflict that it expressly does not have a no-first-use doctrine. With a significantly weaker conventional military, Pakistan uses the threat of nuclear first use as a deterrent. Regarding their nuclear arsenals: India and Pakistan both possess numerous warheads and missiles capable of covering each other's entire territories. As of 2013, Pakistan stated it would need to assemble a nuclear-armed missile from parts stored in different locations, though it's unclear if this remains true. India has new missiles that can be permanently mated with warheads and stored ready for use. It's unclear if any of these missiles are currently mated to nuclear warheads. With that background established, I'm curious where you all are starting on this question. What's your rough estimate of the probability of at least one nuclear weapon being launched by either India or Pakistan in the next 365 days? How are you arriving at that, and what are you considering? Let's start with Nuno. Nuño Sempere 00:02:44 Before this recent escalation, I was at approximately 0.1%. I arrived at that by decomposing the question into: first, the chances of a significant escalation, and second, conditional on that escalation, the chances of a nuke being used. In our forecast a week ago, we defined the threshold for significant escalation as 1,000 deaths. This is much higher than the 27 deaths seen so far. However, India's response seems more kinetic than I expected. If I was previously at a 12% chance of significant escalation and a 1% chance of nuclear use conditional on that, my estimate for hitting that 1,000-death threshold might now be around 15% to 20%. This brings my overall estimate from 0.1% to perhaps 0.2%. Overall, I believe it's still unlikely because both sides really don't want it. However, there's a dynamic where both sides are uncertain, capable of making mistakes, and could start on a path they cannot stop. Rai Sur 00:03:58 Vidur, what about you? Vidur Kapur 00:03:59 There are a few different ways to approach this question. One is, like Nuno, to take this in stages. You could look at the probability of a significant escalation—for example, 1,000 military fatalities within the next 12 months—and then, conditional on that, the probability of nuclear weapon use. For me, I'm now at around 20% for 1,000 military fatalities within the next 12 months. Conditional on that, I'm at possibly around 2% to 3% for actual nuclear weapon use. This puts me a fair bit higher than Nuno, at around 0.4% to 0.6% overall. Taking the average, I estimate a 0.5% chance of nuclear weapon use by either India or Pakistan within the next 12 months. Rai Sur 00:05:11 It might be good to contextualize this second part: where the nuclear use percentages come from. Vidur Kapur 00:05:16 Absolutely. Rai Sur 00:05:17 There seems to be a range of numbers most forecasters coalesce around for nuclear weapon use. How do you arrive at those figures? What's their origin? Vidur Kapur 00:05:29 The second part is to determine the base rate for nuclear weapons use in a conflict. In the past 80 years, coinciding with the 80th anniversary of the end of World War II, there has been essentially one instance of nuclear weapon use in conflict—the two bombs in Japan, which I count as one event. This gives a base rate of 1 in 80, or about 1.25%. Looking closer, when the United States used nuclear weapons, it was the sole nuclear state. Japan couldn't retaliate as it lacked nuclear weapons, possessing only a crude program. The Soviet Union, Germany, and Britain also didn't have them at that time. This suggests the probability of nuclear use in the modern age is lower than 1.25% because states are hesitant to break the norm against their use. In the India-Pakistan case, both have nuclear weapons, meaning mutual use would cause immense destruction on both sides. This might push the probability lower. However, Pakistan's military doctrines suggest a willingness to use nuclear weapons under certain circumstances. One scenario involves a battlefield commander using tactical nuclear weapons, which Pakistan possesses. Doctrines indicate a willingness to use them, especially against an existential threat to Pakistan. One could also imagine a general, in the fog of war, deploying tactical nuclear weapons to halt an Indian advance. If India made an incursion, a general, perhaps misinformed about broader Indian gains, might panic and use them. This specific scenario is concerning and pushes my estimate higher than 1.25%. Generally, the ongoing conflict itself suggests a higher probability. If India and Pakistan were at peace, nuclear use would be unlikely, barring an accident. But with an active conflict and reports of downed Indian jets, we should consider a rate higher than 1.25%, possibly 2% to 3%. We also know that former Secretary of State Mike Pompeo stated in his memoir that during the 2019 India-Pakistan escalation, both countries were very close to using nuclear weapons. While he might have exaggerated for his memoir, it's another data point suggesting these countries are among those most liable to use nuclear weapons in a conflict scenario. Nuño Sempere 00:09:25 Fedor, that 1.25% is for all countries collectively. What share of that risk do you assign specifically to India and Pakistan? Considering the large number of nuclear states, the base rate per year for a state *not* using nuclear weapons is quite high. Conversely, there have been events close to nuclear weapon use. North Korea's nuclear tests to deter the US are an example. More recently, Russia's seizure of the Zaporizhzhia nuclear plant could also be considered, potentially expanding the reference class. What are your thoughts on this, especially regarding the share of the 1.25% attributable to India and Pakistan? Vidur Kapur 00:10:22 My estimate of 0.4% to 0.6% implies a significant share for India and Pakistan. Much of the remaining risk, in my view, currently lies with Russia in Ukraine, as Russia has explicitly discussed using nuclear weapons. In autumn 2022, Western policymakers were very concerned about Russian nuclear use; the British Prime Minister was reportedly monitoring weather patterns due to fears of nuclear fallout. So, Russia and Ukraine account for a large share. However, aside from these two, it's justifiable for India and Pakistan to represent a substantial portion of the nuclear risk within the next year, given their current conflict. As we speak, Indian pilots may be hospitalized due to downed fighter jets. Indian officials have briefed The New York Times and The Hindu that at least three fighter jets are down. While they don't constitute the entire risk, they justifiably form a major part of the nuclear risk for the coming year. Rai Sur 00:11:44 Lisa, what are your thoughts here? Lisa 00:11:46 My forecast for nuclear weapon usage risk aligns closely with Nuno's, around 0.1% to 0.2%. I'll say 0.2% due to recent escalation. However, my overall view is that because both are nuclear powers, the risk of them ever using nuclear weapons is extremely low; the consequences are too high. An incredible escalation would be necessary before we'd see nuclear weapons used. Consider Ukraine versus Russia: despite numerous threats of Russian nuclear use contingent on Ukrainian actions, it hasn't happened. No rational country wants to use nuclear weapons in almost any scenario. Therefore, the risk is extremely small. Most of the current risk, in my opinion, stems from accidents or misunderstandings. For instance, if one country inaccurately perceived the other's actions, it might lead to a misjudgment and nuclear use. This is the most likely path to nuclear weapon use in this conflict. Another factor is the significant imbalance in country size, military size, and military budgets. Pakistan spends $10 billion annually on defense, while India spends $81 billion. Due to this asymmetry and other reasons, neither side wants this conflict to escalate. They will likely tr

    31 min
  • 5 Episodes

About

World-class forecasters make sense of the world and how events could escalate into globally-catastrophic events. blog.sentinel-team.org

Information

  • Creator
    Rai Sur
  • Years Active
    2K
  • Episodes
    5
  • Rating
    Clean
  • Copyright
    © Nuño Sempere
  • Show Website
    The Sentinel Podcast