Glen, I would be more surprised if it wasn't. I flew CSAR for 20 years, left that career 25 years ago, and even THAT long ago, the use of (relatively) automated mission planning software was changing things for the better. Had a couple of follow-on careers (full stop three years ago) and the use of automated Air Tasking Order generation was turning an art into a full-blown science operating at a remarkable speed...several years ago. Given the speed of software iteration, rewrite and replace seen out in the open-source wild, what's happening "behind the black door" can only be guessed at...but it's a good guess there's more than ATO generation going on. A very good guess, IMO. OBTW, the boys (and girls, I'm sure) did good. Really, really good.
When a general purpose, publicly available, free AI/LLM like Grok can take a single sentence from me, search the web for supporting data, outline an essay, fill it in with my “voice”, and present it to me for review in about 3 seconds — yeah, I’m positive that the US military is using some very sophisticated AI tools for mission orders, tasking, planning, etc — if they aren’t, I’d be completely shocked.
It would be interesting to see if AI is behind the decision to ignore the Strait of Hormuz issue. I’m sure they have war gamed the problem but also discovered it was the eastern hemisphere’s problem and not ours. So, make it their issue to resolve (Iranian blackmail) and not ours. Equilibrium will be achieved sooner or later. Then the Eastern Hemisphere can decide if they want their economy to depend on the Iranians. This is a broad statement but with the US having great leverage with the decision.
I was astonished, and still am, at the complexity and ease with which they built a base and found and rescued that pilot, within enemy territory! Our CinC and SecWar are absolutely the best. I have never been prouder or felt more secure as an American. And Trump listens to everybody--I have no doubt his allies in tech world have been advising him a lot on all of this.
I'm just glad those guys are on OUR side and the humans on the ground had skills, great human support in the command structure, the resources, guts and some blessings from the Almighty to do a job well done and get everybody home while kicking the ass of the enemy and making America's critics look really, really, really bad (again!).
Interesting. Similarly, but off-topic, I wonder if they finally managed to shoot down something. Was it just an odd coincidence, or is some of what Biden donated to the Taliban getting out?
In all cases, it was a great job to successfully rescue the wounded airman.
I’d guess it was more of a combination, with lots of different humans using ai & knowledge of their own capabilities. “How to rescue a small group in Iran” is not so far from “How to land and support a small group who are extracting the uranium, after it is found”. Including fighting the defenders of the uranium. There has been lots of planning & preparing to coordinate for the uranium; using those plans to rescue the airman seems most likely to me.
Using ai to better prepare plans, check plans for consistency, and get sign offs from all areas seems an excellent use of ai, available now with the right organizational changes.
My grandson-in-law is the NCOIC of the CENTCOM command post. They do lots of continency planning. Unlike the Carter years they test many of them using real multi-service assets.
What you say is plausible and probably likely. There is a saying that 90% of anything is crap, but I'm not sure if that applies to bureaucracies - the number would be much higher than 90%. AI will greatly reduce the number of people needed to complete a task, so let's hope that, for those who are replaced, their labor is put to better use elsewhere.
Combat Search and Rescue (CSAR) had to be ready to go immediately. They live for the chance to do their job. AI probably could have given instant input on terrain, weather, etc., but not on what to actually do. Great CIA gambit.
agree, the SAR community is always gaming new scenarios and variants of old scenarios; I would be surprised if they did not have a template for this general case. Of course as you point out, many factors would have to be taken into account in real time.
Yes. Way back in the 60s, I used the computer to work on Engineering Solutions. When I started, we used many of same methods we had used by hand. However as computers became faster and we became smarter, we began to solve much more complex problems, and we built models (involving matrix and beyond solutions) so that we could see what happened when every thing happened at once or in very short sequence. There is much artificial intelligence built into weapons systems going back in the 80s. We just never got around to applying it to non-mathematical thinking until now.
A year ago I’d have said that was absurd. From what I understand of the underlying framework of LLMs it still seems grossly unlikely—except that I’ve seen lots of reports of things that are actually claimed as AI output that I’d have said the same thing about — vibe-coding, math and theoretical physics, video generation, etc.
A year ago I bought a month of top-tier ChatGPT to give me cover images for a couple of my late brother-in-law’s books. It took a month of crap rejections and near-misses that I tried to fix with editing tools, but by the end of the month I had a few that I was pretty happy with.
Last month, just for a lark, I dipped my toe back into the *free* version of ChatGPT. I didn’t reuse the old prompts, though I guess I did use some of the knowledge I had gained about what had worked. The very first image it offered me was damned near as good as the best from that whole month. I didn’t try to improve it because I seems vaguely anal to reissue the books just to give them new covers.
A long tangential story, to communicate my moral that things are apparently happening faster and more broadly than I would ever have guessed—and that’s just the stuff that’s public. If you are right, my big fear is that it might convince our government (or the government of, say, China) that it’s now or never to achieve dominance over the rest of the world.
Sounds like they're using AI to do the clerical work. I hae drafted Annex Ks for a variety of command levels, and so much of it is just formatting and typing. Just eliminating that frees up the decision makers to review and decide the stuff that varies from one operation to the next.
Glen, I would be more surprised if it wasn't. I flew CSAR for 20 years, left that career 25 years ago, and even THAT long ago, the use of (relatively) automated mission planning software was changing things for the better. Had a couple of follow-on careers (full stop three years ago) and the use of automated Air Tasking Order generation was turning an art into a full-blown science operating at a remarkable speed...several years ago. Given the speed of software iteration, rewrite and replace seen out in the open-source wild, what's happening "behind the black door" can only be guessed at...but it's a good guess there's more than ATO generation going on. A very good guess, IMO. OBTW, the boys (and girls, I'm sure) did good. Really, really good.
When a general purpose, publicly available, free AI/LLM like Grok can take a single sentence from me, search the web for supporting data, outline an essay, fill it in with my “voice”, and present it to me for review in about 3 seconds — yeah, I’m positive that the US military is using some very sophisticated AI tools for mission orders, tasking, planning, etc — if they aren’t, I’d be completely shocked.
It would be interesting to see if AI is behind the decision to ignore the Strait of Hormuz issue. I’m sure they have war gamed the problem but also discovered it was the eastern hemisphere’s problem and not ours. So, make it their issue to resolve (Iranian blackmail) and not ours. Equilibrium will be achieved sooner or later. Then the Eastern Hemisphere can decide if they want their economy to depend on the Iranians. This is a broad statement but with the US having great leverage with the decision.
Wargaming like this is something I never even considered. Great comment.
I was astonished, and still am, at the complexity and ease with which they built a base and found and rescued that pilot, within enemy territory! Our CinC and SecWar are absolutely the best. I have never been prouder or felt more secure as an American. And Trump listens to everybody--I have no doubt his allies in tech world have been advising him a lot on all of this.
I'm just glad those guys are on OUR side and the humans on the ground had skills, great human support in the command structure, the resources, guts and some blessings from the Almighty to do a job well done and get everybody home while kicking the ass of the enemy and making America's critics look really, really, really bad (again!).
Interesting. Similarly, but off-topic, I wonder if they finally managed to shoot down something. Was it just an odd coincidence, or is some of what Biden donated to the Taliban getting out?
also...Israeli Intel inside...
In all cases, it was a great job to successfully rescue the wounded airman.
I’d guess it was more of a combination, with lots of different humans using ai & knowledge of their own capabilities. “How to rescue a small group in Iran” is not so far from “How to land and support a small group who are extracting the uranium, after it is found”. Including fighting the defenders of the uranium. There has been lots of planning & preparing to coordinate for the uranium; using those plans to rescue the airman seems most likely to me.
Using ai to better prepare plans, check plans for consistency, and get sign offs from all areas seems an excellent use of ai, available now with the right organizational changes.
My grandson-in-law is the NCOIC of the CENTCOM command post. They do lots of continency planning. Unlike the Carter years they test many of them using real multi-service assets.
I'm not surprised at their success.
What you say is plausible and probably likely. There is a saying that 90% of anything is crap, but I'm not sure if that applies to bureaucracies - the number would be much higher than 90%. AI will greatly reduce the number of people needed to complete a task, so let's hope that, for those who are replaced, their labor is put to better use elsewhere.
Combat Search and Rescue (CSAR) had to be ready to go immediately. They live for the chance to do their job. AI probably could have given instant input on terrain, weather, etc., but not on what to actually do. Great CIA gambit.
agree, the SAR community is always gaming new scenarios and variants of old scenarios; I would be surprised if they did not have a template for this general case. Of course as you point out, many factors would have to be taken into account in real time.
Two words: Palantir Gotham
https://www.palantir.com/platforms/gotham/
The lion is known by the claw.
Yes. Way back in the 60s, I used the computer to work on Engineering Solutions. When I started, we used many of same methods we had used by hand. However as computers became faster and we became smarter, we began to solve much more complex problems, and we built models (involving matrix and beyond solutions) so that we could see what happened when every thing happened at once or in very short sequence. There is much artificial intelligence built into weapons systems going back in the 80s. We just never got around to applying it to non-mathematical thinking until now.
A year ago I’d have said that was absurd. From what I understand of the underlying framework of LLMs it still seems grossly unlikely—except that I’ve seen lots of reports of things that are actually claimed as AI output that I’d have said the same thing about — vibe-coding, math and theoretical physics, video generation, etc.
A year ago I bought a month of top-tier ChatGPT to give me cover images for a couple of my late brother-in-law’s books. It took a month of crap rejections and near-misses that I tried to fix with editing tools, but by the end of the month I had a few that I was pretty happy with.
Last month, just for a lark, I dipped my toe back into the *free* version of ChatGPT. I didn’t reuse the old prompts, though I guess I did use some of the knowledge I had gained about what had worked. The very first image it offered me was damned near as good as the best from that whole month. I didn’t try to improve it because I seems vaguely anal to reissue the books just to give them new covers.
A long tangential story, to communicate my moral that things are apparently happening faster and more broadly than I would ever have guessed—and that’s just the stuff that’s public. If you are right, my big fear is that it might convince our government (or the government of, say, China) that it’s now or never to achieve dominance over the rest of the world.
Sounds like they're using AI to do the clerical work. I hae drafted Annex Ks for a variety of command levels, and so much of it is just formatting and typing. Just eliminating that frees up the decision makers to review and decide the stuff that varies from one operation to the next.