AI-Military Integration Under President Trump
Summary
Hamza Chaudhry explores the evolution of AI-military spending, emphasizing the role of structural factors over political principles.
SESSION Transcript
Hi folks. I've been working on a book chapter mapping out the entire history of AI spending at the Pentagon. So I thought it was best to compress this into a five-minute talk at the end of a two-day conference. I was banking on most of you not being here, but here it goes.
I'm going to talk about three things. One, why folks who come to these kinds of conferences should think about this issue. Second, briefly talking about Trump 45 and continuity under President Biden on these issues. And then finally some differences and commonalities under Trump 47.
First, there's a long history of AI spending at the Pentagon. Just a year after the Dartmouth Conference in 1956, you had these projects being set up in particular at MIT and Stanford, but also other places on speech, translation, cryptography, code breaking, which were really ahead of their time.
And then you go through the Shakey the robot phase in the seventies all the way up to the grand challenges in the 2000s. And then finally the third offset strategy which really centers AI as being a core strategic priority. All of this is just to say there's a very long and proud history here of what DoD has been doing on AI and all of what's happening right now is situated in that sort of seven-decade history.
Why should you care about this if you care about AI policy and where government money goes? Most government money related to AI is going to defense and military spending. Depending on how you do the math, it's either 20 times more or less or 50 times more. So that's one reason to think about this.
The second is the fairly bipartisan and consistent positions on AI and military issues, which can be really good because it allows us operational space to make incremental reform. Or it can be bad because it's then hard to move a bipartisan consensus on say things like autonomous weapons. Third, there's a prospect of generating dangerous capabilities in these settings. So that's another reason to care about them.
Fourth, if there's one factor which affects what happens with the CCP and the PLA, it's probably the US government posture on AI military integration. So that's another reason to think about this in terms of global coordination. And then finally over the past year, I'm sure that everyone's heard about this rhetoric around the AI Manhattan Project, which has gotten more pervasive and louder over the past couple of months.
So that's another reason to center the Pentagon when we think about AI spending. What happened under Trump 45? Many, many, many interesting things. So there's projects like Project Maven and Ghost Fleet Overlord. Project Maven is a good example of a project which is focused on machine learning analysis, mostly focused on things like target identification. And Ghost Fleet Overlord is a good example of a project focused on autonomous vehicles, in this case naval vehicles. There were hundreds of projects like these across the Pentagon. What's interesting was the setup of the Joint AI Center, JAIC, at the Pentagon, which became then rolled up into the CDAO under President Biden. Fairly perfect institutional continuity there.
And then second, the National Security Commission on AI, which was set up in 2018, again continues its work largely unabated under President Biden, and then publishes its final report in 2021. The interesting one is also JADC2. So JADC2 is Joint All-Domain Command and Control. The effort here being to connect different combatant commands and domains across the Pentagon into a single AI military whole.
And there's lots of work been going into that since the summer of 2019. You can basically draw a straight line between that work and work that's being done in 2025. So, for instance, under President Biden, the terminology slightly changed. It became CJADC2.
But there's been a huge expansion of those projects at the Air Force, at the Army, at the Navy, and there's many others that I'm not covering here. And then based off of some of President Trump's initial work on setting up marketplaces for AI startups to invest in the Pentagon, you now have a formalized market system called Tradewinds that lots of startups use.
The newer stuff was more to do with what was happening in the world. So Task Force Lima was a task force that was set up to investigate the use of generative AI products at the Pentagon. There was just not a big priority under the first administration because of where the technology was. And then Replicator, which I'm sure many folks have heard about here, this plan to have, like, drone swarms of thousands of autonomous weapons systems is very close to the exact thing that you were hearing from Secretary Mattis back in 2017 and 2018.
Under President Trump, it's been, like, interestingly, more of the same. I don't know how many folks went to these, but last month, once they came out with their, like, you know, AI defense strategy, there were a bunch of events across D.C.
And you could basically map everything that was being said there to what Undersecretary Hicks was saying a year ago. So more or less, it's basically the same, like, an expansion of combatant commands, more investment in autonomy, more investment in startups. The House bill, the Big Beautiful Bill Act, has many, many provisions in it which basically map onto exactly what President Biden's team is pushing. But there's some differences.
There's more Manhattan Project-oriented rhetoric, there's more high-level skepticism about regulating AI developers. There's more proximity to Silicon Valley. My basic thesis is that these developments in AI military integration are not principle-driven, but are driven by structural factors, such as Silicon Valley becoming a more powerful stakeholder, changing its position on AI military issues, concerns of defense modernization becoming more urgent every single day.
And then of course, what's happening in China affects what's happening in D.C. So if there's one thing I want you to take away from this, it's that you should have left five minutes ago. But if there's like one more thing I would want you to take away, it's probably that the future of AI military integration won't depend on principles. President Biden, President Trump are two incredibly different people.
And that makes the continuity on the AI military side even more remarkable. So it's these wider structural factors we have to keep in mind to map that space out. Thank you.
I'm going to talk about three things. One, why folks who come to these kinds of conferences should think about this issue. Second, briefly talking about Trump 45 and continuity under President Biden on these issues. And then finally some differences and commonalities under Trump 47.
First, there's a long history of AI spending at the Pentagon. Just a year after the Dartmouth Conference in 1956, you had these projects being set up in particular at MIT and Stanford, but also other places on speech, translation, cryptography, code breaking, which were really ahead of their time.
And then you go through the Shakey the robot phase in the seventies all the way up to the grand challenges in the 2000s. And then finally the third offset strategy which really centers AI as being a core strategic priority. All of this is just to say there's a very long and proud history here of what DoD has been doing on AI and all of what's happening right now is situated in that sort of seven-decade history.
Why should you care about this if you care about AI policy and where government money goes? Most government money related to AI is going to defense and military spending. Depending on how you do the math, it's either 20 times more or less or 50 times more. So that's one reason to think about this.
The second is the fairly bipartisan and consistent positions on AI and military issues, which can be really good because it allows us operational space to make incremental reform. Or it can be bad because it's then hard to move a bipartisan consensus on say things like autonomous weapons. Third, there's a prospect of generating dangerous capabilities in these settings. So that's another reason to care about them.
Fourth, if there's one factor which affects what happens with the CCP and the PLA, it's probably the US government posture on AI military integration. So that's another reason to think about this in terms of global coordination. And then finally over the past year, I'm sure that everyone's heard about this rhetoric around the AI Manhattan Project, which has gotten more pervasive and louder over the past couple of months.
So that's another reason to center the Pentagon when we think about AI spending. What happened under Trump 45? Many, many, many interesting things. So there's projects like Project Maven and Ghost Fleet Overlord. Project Maven is a good example of a project which is focused on machine learning analysis, mostly focused on things like target identification. And Ghost Fleet Overlord is a good example of a project focused on autonomous vehicles, in this case naval vehicles. There were hundreds of projects like these across the Pentagon. What's interesting was the setup of the Joint AI Center, JAIC, at the Pentagon, which became then rolled up into the CDAO under President Biden. Fairly perfect institutional continuity there.
And then second, the National Security Commission on AI, which was set up in 2018, again continues its work largely unabated under President Biden, and then publishes its final report in 2021. The interesting one is also JADC2. So JADC2 is Joint All-Domain Command and Control. The effort here being to connect different combatant commands and domains across the Pentagon into a single AI military whole.
And there's lots of work been going into that since the summer of 2019. You can basically draw a straight line between that work and work that's being done in 2025. So, for instance, under President Biden, the terminology slightly changed. It became CJADC2.
But there's been a huge expansion of those projects at the Air Force, at the Army, at the Navy, and there's many others that I'm not covering here. And then based off of some of President Trump's initial work on setting up marketplaces for AI startups to invest in the Pentagon, you now have a formalized market system called Tradewinds that lots of startups use.
The newer stuff was more to do with what was happening in the world. So Task Force Lima was a task force that was set up to investigate the use of generative AI products at the Pentagon. There was just not a big priority under the first administration because of where the technology was. And then Replicator, which I'm sure many folks have heard about here, this plan to have, like, drone swarms of thousands of autonomous weapons systems is very close to the exact thing that you were hearing from Secretary Mattis back in 2017 and 2018.
Under President Trump, it's been, like, interestingly, more of the same. I don't know how many folks went to these, but last month, once they came out with their, like, you know, AI defense strategy, there were a bunch of events across D.C.
And you could basically map everything that was being said there to what Undersecretary Hicks was saying a year ago. So more or less, it's basically the same, like, an expansion of combatant commands, more investment in autonomy, more investment in startups. The House bill, the Big Beautiful Bill Act, has many, many provisions in it which basically map onto exactly what President Biden's team is pushing. But there's some differences.
There's more Manhattan Project-oriented rhetoric, there's more high-level skepticism about regulating AI developers. There's more proximity to Silicon Valley. My basic thesis is that these developments in AI military integration are not principle-driven, but are driven by structural factors, such as Silicon Valley becoming a more powerful stakeholder, changing its position on AI military issues, concerns of defense modernization becoming more urgent every single day.
And then of course, what's happening in China affects what's happening in D.C. So if there's one thing I want you to take away from this, it's that you should have left five minutes ago. But if there's like one more thing I would want you to take away, it's probably that the future of AI military integration won't depend on principles. President Biden, President Trump are two incredibly different people.
And that makes the continuity on the AI military side even more remarkable. So it's these wider structural factors we have to keep in mind to map that space out. Thank you.
