Loading summary
Transcript29 lines
- [00:00]
A
Today is Saturday, March 7th. Artificial intelligence is no longer just changing how we work and search online. It's changing how wars are fought. From Ukraine to Gaza to US Strikes against Iran. We are now in an era of AI enabled warfare. So what does that actually mean? And who gets to decide what's allowed when it comes to AI and the military? It's a question that was highlighted just this week in a dispute between the Pentagon and the AI company Anthropic. So our guest today is Paul Shari. He's the executive vice president at the bipartisan think tank the center for a New American Security. He's served multiple tours to Iraq and Afghanistan and is a former defense official who played a leading role in establishing policies on autonomous systems. He was named one of Time magazine's most influential people in AI and has written two books on the topic, first, army of Autonomous Weapons and the Future of War, and his latest, 4 Power in the Age of Artificial Intelligence. So today he's helping explain how AI is already shaping military operations, why the debate over human control is so urgent and important, and how this technology could shift the global balance of power in the years ahead. Welcome to the NEWSWORTHY special edition Saturday when we sit down with a different expert or celebrity every Saturday to talk about something in the news. Don't forget to tune in every Monday through Friday for our regular episodes where we provide all the day's news in less than 15 minutes. I'm Erica Mandy. It's now time for today's special edition Saturday. Paul Shari, thank you so much for joining us here on THE newsworthy.
- [01:33]
B
Thanks for having me.
- [01:34]
A
So in the news lately we've seen the AI company Anthropic be unwilling to compromise on two issues. One, using AI for mass surveillance of Americans and two, using weapons that don't need human involvement, AKA autonomous weapons. Do you expect the military to use AI in one or both of those ways, despite what they might be saying now?
- [01:55]
B
Well, the Pentagon has said that they are not intending to use AI in either of those applications, at least today. Now, there are domestic laws here in the United States that govern the military and what they can do inside the country. And so the military shouldn't be involved in domestic mass surveillance in the first place. Lethal autonomous weapons are a trickier issue. There are no laws in the United States that prohibit them or internationally. There's a policy that the Pentagon has, but it's pretty flexible. And that might be a thing where down the road in a few years as the technology evolves, the military might want to use AI for that. So it's not an immediate issue. But certainly autonomous weapons is one that we're going to face as the technology evolves.
- [02:39]
A
And what are some of the decision points there? You know, obviously there's concern that AI is going to mess up if there's no human oversight. What else do you think people need to be thinking about? Both in that's concerning for it, but also if we don't use it, we could be behind.
- [02:53]
B
Well, I think you sort of actually put your finger on these two twin factors that are driving how people are thinking about autonomous weapons. The first one is, well, what if the AI gets it wrong? We're talking about life and death decisions as the US is in the middle right now of a bombing campaign in Iraq. But as we see AI integrated more. Anyone that's worked with any kind of AI system knows that they can make mistakes and they could screw things up. And used properly, AI might reduce mistakes. But if there are accidents, we could see things be much, much worse. That's a big concern. But the Pentagon is principally concerned about falling behind adversaries because AI technology is widely available globally. Other competitor nations like China are going to have access to AI technology and the military doesn't want to fall behind.
- [03:41]
A
And let me ask about the mass surveillance of Americans because generally that would be against the law. But some people debate on could there be loopholes such as national security reasons and things like that. So is it still something we can't completely write off as happening?
- [03:59]
B
Well, I think one of the things that AI is going to enable is the ability for the government. Frankly, anyone has access to large amounts of data to process that data in a way that has not been fed feasible today. And that raises really novel concerns because AI enables this sort of techno authoritarianism that is not possible beforehand. And we're starting to see elements of this deployed inside China where the Chinese Communist Party is using AI and surveillance technology to monitor and surveil our citizens. And that's not a model we want to be importing here into the United States.
- [04:36]
A
And so what differences, if any, are you seeing between the deal the Pentagon ultimately struck with OpenAI, Anthropic's rival, versus the one that they were not willing to do with Anthropic.
- [04:48]
B
The Pentagon had given Anthropic a deadline, really an ultimatum to agree to the Pentagon's terms, and they did not agree, causing the Pentagon to say that they were going to phase out Anthropic's tools and in fact label the company a supply chain risk. And just a few hours later, OpenAI announced their own deal with the Pentagon to share their AI technology with. Now, OpenAI has shared some details of their contract and then how their technology is being used. We don't know all the details, but some of it may just come down to what the companies are willing to tolerate. Dario Amadeh, the CEO of Anthropic, had said was that the Pentagon had made some movement in their direction, but ultimately the language that they were presented with had a bunch of loopholes, and that wasn't sufficient for Anthropic. I will say one of the silver linings of this incident is it's brought some of these issues to the public discussion because we all have a stake in how the military is using this technology.
- [05:44]
A
Right. We've got the military, we've got tech companies, and then there's lawmakers who, at least for now, don't seem to be keeping up with the technology, which is a common theme we've seen in the technological age.
- [05:56]
B
Yeah, I mean, look, it's kind of trite to say, well, who should be deciding here? Should it be leaders of the Pentagon or should it be tech companies? Well, it really should be democratically elected lawmakers who are accountable to the people. But boy, is there a disconnect between the pace of how AI technology is moving and the dysfunction in Washington. And so the idea that Congress is going to pass legislation that's going to be able to keep pace with AI technology would be great. It's not realistic. Washington really hasn't even caught up to the social media age and some of the challenges that that brings. It's just kind of one piece of this broader problem of the technology is moving much, much faster than our institutions and our ability to govern the technology.
- [06:40]
A
And meanwhile, we're seeing the Pentagon really punish a private company for not going along with what it wants. What do you make of that? Is it based on a particular policy? Is it just to send a message to other companies? What do you think people need to know here?
- [06:55]
B
This is the thing where the whole debate kind of comes off the rails, because it's one thing if they can't agree on the terms of the military doesn't have to use Anthropic's AI tools. Anthropic shouldn't have to sell their technology to the government if they don't agree to the government's terms. What the Pentagon has done in retaliation is launch this really unprecedented campaign of retribution against Anthropic by labeling them a supply chain risk. That's a Designation that's normally reserved for foreign companies. And that basically says that defense contractors have to make sure they're not relying on this company's products as they're fulfilling defense contracts. For the government to punish a private company in this way is really unprecedented and I think is going to have a very damaging, chilling effect.
- [07:41]
A
You know, you brought it up a little bit. The latest conflict happening involving America and Iran. Looking at that, what kind of AI is likely being used now in the current attacks on Iran? And then after that, we'll talk about what could be coming down the line.
- [07:57]
B
It has been publicly reported that Anthropic's AI tools integrated into US Military classified networks were used in Venezuela and are being used in U.S. military operations in Iran. The details of how these are being used, we can make some reasonable guesses. One of the things that AI is really good at is helping to process and understand large amounts of information very quickly. Well, if you look at the US Military strikes against Iran, the US Military's hit over a thousand targets. Now they're having to reprioritize targets, do battle damage assessment on the ones that have been hit. One of the things the military is targeting is Iran's ballistic missile capability. And that's a thing where the ability of AI to help synthesize this information, to go through these target lists, to help prioritize them. It's possible to do it as humans, but doing that at human speed is going to take a really long time. And AI can help accelerate that process. And so that is some of the uses that I think are probably pretty plausible in how the US is using AI in their operations today.
- [09:04]
A
And what else do you think the average American needs to understand about how AI is changing or will change down the line, how the military operates?
- [09:13]
B
Well, I think it's pretty clear that we are now in an era of AI warfare, not just in Iran, but also in Ukraine, in Israel's operations in Gaza. That's going to continue to evolve, particularly as AI technology is improving very, very rapidly. And because the technology is so widely available around the world, militaries are going to be very hungry to find ways to speed up their operations, be more effective, be more precise. But there are real risks. And fundamentally, the question is that we are gonna face how do we maintain human control over warfare that's being fought at machine speed? Can people keep up even if humans are nominally in control? Humans are approving targets. We're not quite there yet, but that is certainly coming, and it's gonna be a very challenging issue. To address because we need to keep human control over warfare. But militaries are gonna be worried about falling behind competitors. We need to find ways for Silicon Valley and Washington to work effectively together. I think it's healthy that we're having this debate in public because we all have a stake in how this technology is used.
- [10:23]
A
And it sounds like you want to find the balance that one extreme or the other is not going to work out well.
- [10:30]
B
There are lots of ways in which AI has limitations. It makes mistakes. Large language models tend to hallucinate things. They can have strange biases. They might be giving information to feed into the perceptions that humans already have. And I think it's vitally important that we maintain human control over lethal decision making in war. We want to find ways to use the technology to be more precise, to be more humane, to reduce civilian casualties, but not lose our humanity in the process.
- [11:04]
A
Still ahead, why AI could reshape the global balance of power and what the term sovereign AI is really all about. Plus what role all Americans can play in this debate. But first, a quick break for our sponsors. I have enough decisions to make every day that I don't want to decide between a million things in my closet when I'm just trying to get dressed. Instead, I'm now focused on having versatile, well made pieces that make getting dressed simple. And several of those items in my closet are from quints, from a soft cashmere sweater to a washable silk tank that somehow works year round to my diamond bracelet that's great for daily wear. Quince makes high quality wardrobe staples using premium fabrics and now is the time to check out their seasonal colors and prints for spring too. Plus, I appreciate that Quince works directly with safe, ethical factories and cuts out the middlemen so you're not paying for brand markup or fancy retail stores. You don't need a closet full of options. You need pieces that work. And right now go to quince.com newsworthy for free shipping and 365 day returns. That's a full year to wear it and love it. And you will. Now available in Canada too. Don't keep settling for clothes that don't last. Go to Q U I n c e.com newsworthy for free shipping and 365 day returns. Quince.com newsworthy the newsworthy is also brought to you by Oneskin. Look on my bathroom counter right now and you'll find at least three OneSkin products. The face moisturizer, the eye cream and the lip mask. Three of my favorite products that I use both morning and night every day. And I love how all three of them feel on my skin as well as how they make my skin feel and look and how easy and simple the routine is so I can stay consistent. It's not just me either. Oneskin has more than 10,000 5 star reviews where people consistently mention smoother, firmer, healthier looking skin and how easily these products fit into their daily routines. Editors have named Oneskin a leader in skin longevity with recognition from Fast Company, mindbody Green, Bloomberg and the Today Show. Born from over a decade of longevity research, OneSkin's OS1 peptide is proven to target the visible signs of aging, helping you unlock your healthiest skin now and as you age and for a limited time, Try OneSkin with 15% off using code NEWSWORTHY at OneSkin CO. NEWSWORTHY, that's 15% off at OneSkin Co with the code Newsworthy. After your purchase, they'll ask you where you heard about them. So please support our show and tell them we sent you. Now back to my conversation with Paul Shari. You've said in your latest book that you expect profound disruptions in the balance of global power because of AI. Can you explain a little more about what that looks like or how this might play out with other countries that are also trying to develop AI as quickly as possible?
- [13:46]
B
One of the interesting features about AI is that it can be used in a whole variety of different kinds of applications, and so it's similar to other general purpose technologies like electricity or the internal combustion engine. Many people have compared AI to another industrial revolution. Well, if you look at what happened during the Industrial Revolution, nations rose and fell on the global stage based on how rapidly they industrialized. Coal and steel became key inputs of national power. Oil became a geostrategic resource that countries were willing to fight wars over. So what are these key new elements of power in an age of AI? And who's going to win? I think one of the things we're seeing right now is that AI is incredibly computationally intensive. That's why we see this massive build out of data centers, this fierce competition for companies trying to get a hold of the most advanced chips that are coming out. That has become a source of geopolitical tension between the United States and China, because the most advanced ships in the world are actually not built here in the US they're built in Taiwan. This is an island 100 miles off the coast of China that the Chinese Communist Party has pledged to absorb by force if necessary. But those fabs in Taiwan depend on us Technology. And so the US has put in place export controls on where those chips can go and what the threshold is for setting these export controls on the most advanced AI chips to China has been a fierce topic of debate over the last couple years. And it is probably one of the strongest points of leverage that the US Government has to maintain a lead over China in AI. Finding a way to give the United States even a small edge in this competition could have big effects in the long run in terms of economic productivity and in terms of national power.
- [15:38]
A
That's interesting because we also heard the concept of sovereign AI as countries seek to kind of reduce their dependency on other countries for their technology for AI. Can you speak to that a little bit and potential importance of it and how the US Is thinking about it?
- [15:55]
B
Yeah, sovereignty has all the buzz globally right now. So part of this is in response to the very strong leadership position that the US has right now. And the most advanced AI models are coming out of US companies. Chinese companies are not far behind, in part because they're riding on the coattails of US Advancement, like distilling US Publicly available models and extracting data out of them. If you go beneath that, you've got the chips themselves, which are made in Taiwan. But one of the leading chip designers, Nvidia, is a US Company. The data centers that are being built out, the leading cloud providers, are U.S. companies. And so a lot of countries are looking at this, and particularly in light of all the trade wars that the Trump administration has launched globally, even close US Allies and partners are rethinking their dependence on US Technology. And so a lot of countries are saying, hey, we want sovereign AI. We want our own AI models, our own data centers in our own country. Their ability to do that is going to depend both on that country's technical abilities and the wealth that they have available. So countries probably can put their own applications on top of US Models. They could build data centers in their own countries, they could train their own models. Although they might be lagging some of the leading US Companies in areas like chip fabrication, it's really just not feasible. But a lot of countries are trying to find ways to create some sovereign space for them to have their own AI technology.
- [17:28]
A
What isn't getting enough attention, in your opinion, or something that you think is misunderstood that you want to clear up?
- [17:34]
B
I think an important aspect is acknowledging that the debates around autonomous weapons that were one of the key sticking points between Anthropic and the Pentagon are not, as far as we know, at least, an issue in terms of how the military is using technology today, but it is coming, and that's the direction of the technology. And so now is the time to have this conversation. And the core of this debate is about what are the rules for AI warfare, but also who gets to set those rules. I think that public perception of AI is shifting quickly. The conversations that we're having as a country about the effect that AI has on our society, on our jobs, on our welfare, are really important to be having because it's too important to live up to just the government or just private companies who are ultimately motivated by profit or just the two of them alone. We all need to be involved in these conversations. Ultimately, we need to find ways to adopt AI in our society that benefit humans, that don't benefit AI, or just make money for people, but that are benefiting all of us.
- [18:44]
A
So for the listener who cares about this issue, what do you recommend that they can do to get involved in this conversation and to have some influence?
- [18:52]
B
Yeah, I mean, I think there's a variety of things, you know, learn more, be educated, participate in discussions in your community about AI and how it's adopted. Certainly there are communities around the country that are involved in debates about things like data centers that are being built in their backyard. Write your political leadership. Politicians are actually super responsive to feedback from constituents. I mean, I work at a bipartisan think tank in Washington, and I will get outreach from members of Congress saying, hey, I'm getting this flood of letters from my constituents about some topic. And sometimes members of Congress will say, tell me about this, help me understand, and what's the right answer here? Those democratic processes, they're messy. There's a lot of friction. It's easy sometimes to be cynical, but those avenues of leverage from the public do work and are effective, and politicians do respond to that.
- [19:54]
A
Thank you to Paul Shari for sharing his knowledge with us today. We'll link to both of his books if you want to learn more about AI's role in the military and global power. Again, his latest book is called 4 Power in the Age of Artificial Intelligence. And join us again during the week to stay updated on all the day's news. We provide an overview of a wide variety of stories every Monday through Friday in less than 15 minutes a day. So we'll be back on Monday with the latest news to know. Until then, have a great rest of your weekend.