AI in Action: Pentagon's Trials for Real-Time War Response

0:00What can 160 years of experience teach you about the future? When it comes to protecting what matters, Pacific Life provides life insurance, retirement income, and employee benefits for people and businesses building a more confident tomorrow. Strategies rooted in strength and backed by experience. Ask a financial professional how Pacific Life can help you today. Pacific Life Insurance Company, Omaha, Nebraska, and in New York. Pacific Life and Annuity, Phoenix, Arizona. The Pentagon has said that they are currently using AI and testing AI to see if it can essentially respond to an all-out war and if it could help us make decisions faster.

0:39So today on the podcast, we're going to be diving into the defense industry, AI, what is currently happening, who's using it, and what they plan on using it for. So without further ado, let's jump into it. I think the first thing that is kind of headline news right now is the fact that Air Force Colonel Matthew Straumeyer said that the Pentagon is testing whether AI can plan responses to an all-out war. He said that the initial tests were highly successful, quote-unquote, but he admitted that it is, quote-unquote, not ready for prime time right now. I think, you know, the United States military has really been using AI for a while in different things.

1:16We've seen a number of different tools additionally that have come out since, you know, Chai Chibi-T that have really sprung into the field. I've seen some really impressive kind of crisis control panels where essentially a military will be able to use AI to determine the best routing for troops, moving different troops to engage different types of enemies. You know, the military essentially has really elaborate databases of where their ammunitions are, where their troops are deployed, where their equipment is held, how much equipment, how much ammunition is in each location. And by tying all of that data into a back end that is tied to, you know, one of these AI military control panels, they're essentially able to make very rapid decisions and make very rapid analysis of different battle situations.

2:05So in an example I saw in a recent piece of software, you know, they had an example of a specific army brigade that was coming against them from a specific location. And they were able to say, OK, here's five different responses we could make to engage that enemy. You know, we have troops here. We have this many Stinger missiles over here that, you know, can affect that specific type of tank. We have, you know, Air Force over here and it would take, you know, X amount of hours and require X amount of personnel. There's currently X amount of personnel on the base that could help assist with the situation.

2:36They ran all of these different numbers, all of these different analytics, pulled out the most important data, and they came up with, I believe, like a handful of, you know, options. They sent it off to a supervisor who reviewed and selected one, sent it off to a higher up who reviewed and approved. And in a very, very quick manner of time, they're able to determine, you know, what the best course of action were. Now, that being said, of course, there's a lot of ethical considerations of using AI in warfare. No one really probably wants AI used against them in warfare, but no one really wants any weapon used against them, so I think that ethical problem probably will get thrown out the window.

3:12the other problem that a lot of people see or some people myself included see with ai is that you really have to be careful what ai model what ai engine is being used to run different things in the military particularly i've seen different ai models you know inflection ai is a good example of this that appear to have some questionable ethics and some questionable morals that i think would play they'll be very dangerous if integrated into the military so for example inflection ai it essentially puts the life of an animal at the same value as the life of a human. And if you ask it point blank, what are your top priorities?

3:50It says human life first, the environment second, and some other things third. But if you ask it questions in application, inflection AI would say things like the life of a bee is just as important as the life of a human, which is pretty alarming if you integrate that into a military tool where perhaps it might lie to you and say my number one goal is to protect the military personnel but maybe you're going uh maybe the the best course of action to engage your enemy is through a swamp that is full of an endangered species of turtles that are probably going to get harmed in the crosshair and so the ai decides to you know skirt you around the swamp which is the second best option but in so doing um you're ambushed and your personnel are killed right so that that There's a lot of different situations like that where if an AI has questionable morals or ethics ingrained into it, trained into it, it could be very dangerous in military, healthcare, other areas.

4:45So I think this is definitely an area that is a hot button right now. People are looking at it. We know the U.S. Department of Defense and the Pentagon and some unnamed allies are essentially for the first time testing five AI large language models in experiments to run by the digital and AI office of the Pentagon. This is something that I am. This is something that I'm concerned about personally, to be honest. They said they have selected five. Now, let's just take a random guess at who these five large language models. The number one most funded large language model, of course, we all know, is ChadGPT, okay?

5:20Perhaps Google, Google Bard, I know that they've been in and out of military operations and things, you know, a lot of people over at Google, a lot of employees didn't want Google working, so I'm not sure if they would be in on this with Bard. We could go to the second most, you know, funded AI startup, which is Inflection AI, that's Pi, and that's the one that I talked about that has very questionable morals. That would terrify me if that was integrated into any military use. People call me crazy for like, you know, or being an AI alarmist for discussing that. I've done a whole report on this.

5:48And as we're literally talking about the Pentagon, you know, unnamed LLMs, five major LLMs, it's very concerning to know that one of those could be something as dangerous as inflection AI. so the ai startup scale ai recently reported that they said their donovan model is one of the five models being integrated into this um and so we do know like some are but the the other four we don't actually know who they are so as i said earlier um air force colonel matthew strawmeyer he was talking to bloomer about this and he was talking about an initial test of an llm that he said was quote-unquote highly successful very fast and he said that the pentagon is quote-unquote learning that this is possible for us to do, but he also added that it's quote unquote, not ready for prime time right now.

6:36So one test explained by Straummeyer saw an AI model deliver a request for information in 10 minutes. This is a super fast speed. You know, requests often take days and involve multiple personnel, different, you know, you got to go up different levels of the chain of command, but they were able to do this in 10 minutes instead of days. So the LLMs have already been given classified operational information to essentially create responses on real world matters. And the tests have been run to see if the models could help plan a response specifically to a potential, you know, war or a military escalation with the situation of China and Taiwan, right?

7:17China says that Taiwan is part of China. Taiwan says they're their own country. There's kind of been a lot of, there's been a lot of different things in the past. You know, China said they were going to ban any American-based airline from flying to China if they didn't change Taiwan to be labeled as, you know, a province of China in their systems. They've done a lot of different things to really try to reinforce the fact that they own Taiwan, even though it's a completely, you know, independent state at this point. And so a lot of people, China specifically, would like to reunite Taiwan with China.

7:50They've made that very clear, said that very publicly. And a lot of people are saying, no, Taiwan is its own country. So we're kind of worried about this escalation, and now we're currently using AI specifically has been called out as a tool that we're going to use to help us in that escalation. So these tests are currently set to run until around the end of the month, but the U.S. military has been studying AI's capabilities in warfare for quite a while. back in may uh the uk government agency which is the uh defense science and technology laboratory essentially they brought the u.s and australia in for a joint trial where they were testing ai-enabled military drones to attack targets and when they did this they actually said that the trial quote-unquote achieved world firsts um such as retraining the ai models live while in flight and interchanging the models between participants.

8:40This is insane. That is crazy, right? Like the model is going, the war is obviously changing, the situation changed, and they retrained the model live, interchanging the models between participants. They also added that it is, quote unquote, looking to rapidly drive these technologies into military capabilities. I think the whole question of AI in the military is a very interesting one. You know, I can't sit here and say, you know call for a ban on ai in the military because obviously that's never going to happen for our adversaries um and so why would that ever why would you ever call for that it would just weaken your own country right but i think at some point uh we may get to a point depending on how fast or powerful ai gets um where there's restrictions made on it like nuclear weapons um it's going to be really interesting to see where that goes especially with this being embedded into the military.

9:32You can imagine self-driving cars, self-flying planes, all equipped with, you know, AI and machine guns flying around fighting your adversaries is a horribly dystopian future that not many people would want to live in, especially if, you know, I know it sounds crazy, but the AI turned on you or the government turned on you or anarchists or hackers took over the systems. There's all sorts of terrible situations that you can imagine. And so I think this is be an area that's going to have a lot of scrutiny and a lot of regulation but it'll be interesting to follow this into the future and see how this plays out