One advantage of a UCAV that didn't appear to be talked about was how different the airframe would be. You are talking about vehicles that could pull off maneuvers that would kill a human pilot. They also might save weight by cutting things like life support features, control panels, and all the safety features. In theory this should lead to a more lethal air frame with fewer design constraints.
The design differences between the F22/F35 and the current unmanned fighters are drastic visually. They are smaller, flatter, more stealthy. Some are tailless (similar to the B2 and B21). No cockpit, no ejection system, no canopy as visibility is unnecessary. No manual controls. No oxygen necessary. No titanium or better material bathtub.
Incredible content! Paco was outstanding with clear and concise explanations. He has made major improvements himself as he described the complex evolution of electronic warfare and the force multipliers that can be brought upon current adversaries. Absolutely riveting. Kudos to both of you for this presentation.
It made me chuckle to hear Pako say that he wanted a Moochie this year because as I was watching through this video, the one recurring thought I had was "this is one of the best interviews I've ever seen on this channel". Extremely informative. And I say that as someone who normally thinks that Bronk's interviews are just superior to everything else. Not only was this extremely informative...and on a very interesting, very important topic...but I'd say that Pako has gotten more comfortable doing these types of video interviews as well. This one seemed very natural. This was just a truly great interview all around.
With AI, w/o human in the loop, it's not checks and balances (there aren't any after deployment), it's do we understand the risks. I'm doubtful we do. But I think we'll probably deploy until the AI accidentally takes down an air liner and reassess.
@@PatrickProctor-Brown-xf6nd my prediction is that enemy shoots on purpose a massive missile into opponent's big oversea base, claims it was AI that shot accidentally to evade responsibility. but you cant see the insides because its military top secret. only their modified report.
Yea, as far as we know nuke-tipped BOMARCs are the only nuclear weapons Canada ever based on our soil. Nuclear SAMs lol. I shudder to think what would happen if we tried that today, with society so reliant on personal electronics.
I have been wanting to leave a comment for some time, but for the most part I watch videos on the TV so I don't have the opportunity to comment. I truly enjoy your videos and find you to be a VERY reliable and authentic source of information. I REALLY enjoy the episode where you have Paco on, he is fantastic! Keep up the great work! And thank you!!
As a retired researcher/teacher of AI, I want to say that it was a joy to watch Pako, simply because he was spot-on with every single thing he said. (Almost 30 years ago, I worked with Mica Endsley and a grad student on training artificial neural networks to emulate the decisions of expert pilots in air combat simulations. Mica, a major contributor to situation-awareness assessment, went on to serve as Chief Scientist of the Air Force.) I am deeply distressed by the volume of BS about AI I see in the mainstream media, especially at the New York Times. It genuinely brought a big grin to my face, seeing Pako nail the particular topic, and also the general state of affairs in AI, as he did.
Indeed, I can no longer hang out with my buddies when they head out for major beer consumption. The talk turns to speculation about where all this Ayy-Eye might be leading us, reeling off a load of B-movie cliches that were bad science fiction 50 years ago. I end up comforting them that evidently, saturation beer bombing would exterminate any sign of superior intelligence within 50 km.
*It takes 8 hours approximately for a human to read 100 pages of book, it takes days and weeks for AI to read the same because both have profoundly different way of processing and learning.*
@Only_God_Is_Allah_SWT Not recently but not long ago Public Broadcasting Service was interviewing folks who said, "We are now getting answers to questions we hadn't thought to ask. So. Maybe AI must ponder character when reading some kinds of work.
Garmin now makes a general aviation autopilot system should the pilot become incapacitated. There is a button on the center cockpit panel that the right seat passenger can push to activate. The airplane will then go into emergency mode and land at the nearest feasible airport. The system takes into consideration the aircraft's position, fuel, winds aloft and nearest available useful runway and prepares to land. The radio will transmit an emergency transmission on 121.5 mh and the transponder will squawk 7700. The intercom will instruct the passenger(s) what is happening and what to expect. It can do this in IFR conditions too. Garmin even advertises this in popular aviation magazines.
I'm sure they'll have downloaded offline GPS maps that are new, plus terrain following, not to mention the DoD has an emergency satellite they can launch that'll work for 24 hours, I think? Maybe it was 48, I'm sure when I read that it's gotten better. But that's pretty dammed cool regardless, I wonder if that last plane crash that was heading to DC into restricted air space where people heard the sonic boom a month or two ago where the fighters were trying to figure out wtf was going on and got in front to deploy flares to see if they were alive or awake cause it did do a u-turn and went back towards too what appeared where it took off from but crashed, maybe it ran out of fuel since it wasn't supposed to go very far, it was a weird scenario regardless. Mooch did talk and show on a flight record of it, then again I suppose if it had that program in it, it would've just landed in New York where it was originally going, I think? Hell, I dunno that one was freaky, plus it was the airplane CEOs family as well? Really freaky. Must've lost cabin pressure for both pilots to be knocked out. I wouldn't be surprised if the military haven't talked to them about this or already have their own backup GPS system in space that's for their own use if something like this comes about. It's pretty cool technology regardless.
Great share. There is so much talk about AI right now. Being almost 65 and watching some of the movies we watched as kids how can I not be scared of AI? “Do you want to play a game” sticks deeply in my mind. Thank you again for sharing.
AI represents potentially great power. And power can be used in helpful or harmful ways. So maybe rather than being scared a responsible response is to be very cautious. Which the military seems to be doing, but how long that will last is another.
@@davidsmith8997 how can you be cautious instead of scared when you/I have no control over any of what will be done with it? As well as different militaries all over the world utilizing it with no possible way for you/I to know what they are doing with it?
At 80, I have thought about AI for 65 years now, and I am not scared of it, in fact, I use it in my Tesla with FSD. (The wife is a little scared, but she is learning). A great time to be alive. I am a Vietnam War vet and amateur radio operator, WA7VQR with 50 years computer experience. Edit; I think my wife is more scared of my driving than the car's driving.
Here's my vote for Pako's "Moochie", and my ongoing thanks for again presenting such an engaging, important and rapidly changing topic in the now familiar and welcome WC "dueling banjos" (or is that guitars?) format. The two of you work so well together, and contribute something that always adds up greater than the sum of the parts, and any government body seeking a similar update would do well to consider bringing you guys in pronto. Just as a video with two authors (think Shattered Sword) provides an amazing and captivating dual perspective, each of you has invaluable experience to bring to such a gathering of government officials . . . and one that we hope you will be bringing back on home here again soon.
This episode should be nominated for a Streamy Award under the Subject Award category. One of _the_ best and most interesting discussions where you don't need a ton of visual clues to understand and be engaged with the dynamics of the subject matter. My favorite Ward Carroll interview to-date, alongside with Kim "KC" Campbell's episode.
I love Pako's ' The Merge'...great info across a spectrum of topics I follow. Also excellent discussion on the varieties of AI...which is mostly lost in the media and in the public domain.
Love the Tshirt Ward. It’s interesting how AI changes risk profiles of missions and makes me wonder what super risky missions would entail (is this an escalating self reinforcing process?) as well as how adversary AIs will respond to each other in massive swarms.
I think wars fought 70 years from now will basically be video games --- you'll hear a blurb on TV about "today U.S. drones shot down 50 Russian drones" and no human lives lost --- wars will actually be FUN but expensive LOL
@@fredwerza3478 there’s an episode of the original Star Trek like this except the losing side has to exterminate a set number of its own citizens at the end of the computer war game.
Upon your suggestion, I watched Unknown, Killer Robots. A good piece that helps you to to see the ying & yang of it all. That said, the brilliance of the technology staggers me! Thanks for bringing Mike on the show.
Great conversation. Paco was really able to define the components and the roadmap of this technology, clearly and succinctly. Perception -> Cognition -> Action being the phases. Also a well said explanation of AI and machine learning which is so often clouded in buzz words by the public. Complex processes which are deterministic and happen the same each time, by instructions are not AI. But heuristic decision making, where choices are made with probabilities and based on learning from countless past data points is Machine learning.
Fascinating banter with two incredible pilots! Great to be a grunt fly on the wall and listen to a whole new world of war! And the comments from equally qualified people are great! Thank you!
Well, plus they can pull as many g's as the airframe can handle, you save weight since no life support or even just seating are needed, plus the ship can be much smaller since you don't have to fit a man or men inside. Hey Ward, remember the HiMAT test drone?
@@Carlos27thFS Think you need to rethink or just initially THINK that comment. Let's just help you out by breaking it down, shall we, like sounding out unfamiliar words in kindergarten. Trad combat platforms have both pilots and fuel ..... yes. Fuel is usually but not always in the wings. UCAV needs fuel but can completely dispense with the volume and mass of pilot LS and control systems. How does this equate to "gonna use those spaces for fuel". There is no need to use those spaces for fuel. Thus, those craft gain a significant performance advantage, in both thrust to weight and max G limits. If there are aspects we are missing, please explain in detail.
@@jamesburns2232 And young female warfighters have this shocking habit of getting knocked up! And then having to take extended leave or even full separation form service. Who could have expected such a thing to happen? After all there are REGS to stop that sort of thing from happening! (Tongue firmly in cheek).
X-47, XQ-58 Valkyrie, MQ-25 Stingray and the AI F-16s are providing tons of valuable data. Can't wait for the Loyal wingman & Ghost bat to go into service.
As an ace combat veteran, i know how difficult it is to go up against ai combatants. They can pull maneuvers far beyond what is capable for any human, the only way to defeat them in air combat is to be the main character. May god have mercy on our souls.
@@thystaff742Lol they wouldn’t be allowed to do anything to it’s homeland people… That’s silly. We still have yet to have actual “ai” this last 5 generations don’t even know what Artificial Intelligence actually even is, we aren’t even close to getting there. All we can do is program mls. That’s IT!
It never occurred to me that the drone have situational awareness of the complete battle and probably know all the specs of its opponent too, and can make decisions based on all those data-points every few milliseconds.
The human brain can only process so much information per second --- a drone can have many brains and thousands of sensors --- a superior battle machine
But it doesnt have full situational awareness. That is hardware based. Does it have passive/active radar that is 360 degrees around, along with dual mode thermal imaging and uv imaging? Does it also have low light passive imaging 360 degrees? Has it been updated by all the supporting services as to all threats? If the answer is "NO", then there is no such thing as "full situational awareness". The F35 almost does, but even so, there is a ton of work to determine what to do, where, and when. Using a UCAV in conjunction with a pilot in say an F35, that would be about as good as it gets, but it isnt impervious to various forms of jamming or interference.
Well at least it has occurred to you, so far it hasn’t occurred to the US Air Force and the US Navy, because those idiots are still planning on designing airplanes with pilots in them for the foreseeable future. What a joke!
We can already see that the ratio -- of (A) hand-wringing about AI to (B) the unintended harm actually caused by AI in the real world -- is so huge that it borders on infinity.
Until it actually backfires and kills people accidentally once. Then mankind will assign infinite value to whatever harm was done. Small probability, but intense measure if/when it does happen.
@@texasbeast239 that's why autonomous cars are taking so long to hit the road --- one death caused by AI will overshadow the thousands of deaths caused by distracted humans
You mean like people killed in Tesla with the automated driver? You do understand that the "AI" used, is the same thing as the Beoing 737 max series were using , right? AI is only a machine operating a program. Nothing more.
I remember a stars and stripes episode from 2010ish regarding a few pilots going toe to toe with AI counterparts in some simulations, and they were shocked by how good it was. Glad to see this finally came into reality
@@surelyyoujokemeinfailure7531 Ghost Bat can operate in a swarm as is the name for it. Ghost Bat was named after a Australian animal that hunts it's prey in packs using their high sensors. Ghost Bat drone can do the same and why is most advanced AI drone as Australia builds good AI with swarm drone capabilities.
Paco is the best. I wish y’all could have talked longer. He said he couldn’t comment outside of dogfighting, but I believe the AI for BVR air-to-air and all of the components of electronic/jamming/SAMs warfare is where the bread will be buttered. Superior (AI) BVR will allow air-to-air dominance from distance. And electronic/jamming/SAMs (takeout) warfare will allow penetration of enemy defenses and space.
The drones could also carry smaller self guiding drones over the target area and deploy them in such a way they would be difficult to both detect and counter against. They could in essence be programmed to attach to either vehicles of personnel at a distinct asset level.
One threat we will have to contend with from foreign AI adversaries is that they may lack the constraints we put on ours (they may not care if they shoot down a few of their own pilots as long as they get enough of ours). That creates a potential hole in our defenses.
As soon as I saw this I thought of the A.I. battle bots. The battles are over in seconds, and they are getting faster and faster. Also it scares me because, law enforcement moving towards A.I. might be the worst thing I can think of.
Unfortunately I think law enforcement is "all in" on AI --- it's gonna be too easy to use deadly force on people and then have local governments claim legal immunity from it because it wasn't a human pulling the trigger
There is absolutely no reason for AI to keep humanity around. That is the heart of the problem. AI will only keep us around until it figures out how to duplicate our ability to truly have a consciousness and useful creative thoughts. After that, we are done.
Science fiction. All AI is is mountains of probability calculations. That’s all it’s doing. Taking an input and finding the best answer based on statistical calculation. It doesnt feel, desire, or have consciousness.
Fascinating interview Mooch. Your question at the beginning, "Will we have pilots in airplanes in the future" wasn't addressed. F-15E Strike Eagle Weapons System Officer Mike "Pako" Benitez provided some amazing unclassified illustrations and is unable to address Darpa classified issues. Appreciate your and Mike "Pako" Benitez's sides of the AI story.
They will probably end up taking missions with less risk reward than you would only normally take if you were losing. So data will show "wow, we would have lost X number of pilots." And that will be a thing by manufacturers and their supporting high rank officers. What we will never see are the billions and billions wasted on high risk missions that should have never been taken because there was no human life at risk, but chance off total loss was high compared to mission goal or reward.
Pako speaks the truth. My mil days are long behind me but my current career in software is deep in AI. Machine Learning is the major thinking component of AI and his description is spot on. It learns through repetitive events, and we sim thousands of events per day to 'teach' the AI how to analyze and react. The trick will be having enough variability of events to give the AI broad enough training to deal appropriately with all eventualities. SEAD seems like an area for early opportunities for success with AI. Instead of 'Wild Weasel' when humans are involved that will be more like MILD Weasel from a human safety standpoint.
Very good discussion. You're probably right... highest automation is expected. However, as we remove humans from danger, the human's natural desire to avoid the fight (start the war) may to down. It may lead to more war, not fewer deaths. Secondly, how do you cause the aggressor to capitulate if he is only losing "things" not lives.
It would be interesting to see how useful AI is when its sensors are degraded in a realistic combat scenario, jamming, direct energy weapons burning optical sensors, cyber warfare like introduction of spoofed gps data etc, supposedly humans traditionally have the advantage in this environment as they can sort the chaff from the wheat, wonder how robust these agents are when faced with that.
Jamming should be a part of the past. Ultrawideband technology is still evolving - I was alerted to it by one of our communications engineers more than 20 years ago and the FCC is still catching up. The FCC is only concerned with civilian uses in the US, and only in regulated bandwidth. UWB in its advanced forms is virtually impossible to intercept or jam unless the "carrier" pattern of pulses is leaked. The bandwidth of such services is almost literally 3 GHz to daylight and receivers don't care about anything that does not match the "carrier" pattern. Better yet, the extremely low average power requirements mean almost unlimited service years or human lifetimes - without draining the power source.
@@flagmichael Only in bursts is it nearly impossible to detect. Dont forget, every time you intercept a Russian Elint aircraft, that is literally what they are doing. Watching, listening, learning. Every time you communicate or pass messages. It doesnt need to jam every frequency. It only needs to know which ones you use a lot. The whole idea of Growler type aircraft is either to jam specific frequencies, or to share jamming of all possible frequencies.
The first time a UCAV shoots down a manned combat aircraft, it will set a tone for this new arms race. It bypasses some barriers to technology-sharing because we guard certain manned combat technologies from FMS. Imagine UCAV operating in Ukraine right now, for example, especially with a cooperative OCA/strike mission set.
It was attempted before when an MQ-1 Predator drone that had a stinger launcher loaded tried to take a pot shot at an Iraqi MiG-25. The Foxbat fired an R-40 in response and the R-40 had a large enough exhaust that it distracted the Stinger causing it to miss. The R-40 missile then splashed the MQ-1 Predator. First ever engagement between a UCAV and a manned fighter turned out to be a defeat for both the USAF and UCAVs. That said an MQ-1 was probably not the best choice of UCAVs to use against a Foxbat, likewise a Stinger not the best weapon to use either. As for Ukraine I have seen video of a drone engaging another drone. One of their FPV operated drones with a bomb attached to it crashed into a quadcopter the Russians were using for reconnaissance.
@@NuclearFalcon146 MQ-1 isn’t a UCAV though. UCAV is autonomous, jet-propelled, LO/VLO, with a heavier payload and higher speeds. When a UCAV launches an AIM-120C7 through D3 at another fighter from unobserved parameters and splashes that fighter, it’s going to be a game-changer. Same for a transport plane or AWACS, which are even bigger trophies. It will also be a big deal when one or more of them conduct a deeper penetration strike mission with JDAMs.
@@LRRPFco52 UCAV = Unmanned Combat Air Vehicle. It includes remote operated and autonomous. MQ-1s, MQ-9s, and even the Ukrainian quadcopters rigged to drop grenades count.
@@NuclearFalcon146 I'm referring to the UCAV program constantly whispered about dating back to when we were on Global Hawk. UAS is probably the term you're looking for.
Great talk! The explanation of reinforcement learning was great! There’s a lot of awesome AI stuff going on in gaming. Theres an AI project for a team based game that is quite complex. The AI players can work collaboratively with human players, and keep situational awareness of the whole map while in an intense frenetic competition. Something that is incredible to watch, and no player even the worlds best can match. I think one of the big uplifts will be that not only are the airframes simple and relatively cheap, but more airframes in the air at once is a multiplier. With newer radar etc there is a network effect, in addition to the redundancy and capacity of higher airframe numbers.
We should embrace these technologies like these because honestly, we have no choice, they are coming and whoever doesnt have it will be the little brother at the dinner table.
@@glennac Right. And as I've said elsewhere, we've agreed to limit chemical and biological weapons. We've agreed to not research human cloning and some types of genetic research. Seems like the sane thing to do would be to agree to something similar with AI. Anyone caught breaking the rules would be treated like someone caught developing weaponized smallpox or genetically engineering soldiers.
What if China camouflages their drones with fuzzy letters and pictures of crosswalks? Will a 6th gen. WSO be necessary to help the AI pass all the captcha tests?
seems like having an ultra-high ceiling, insane speed control craft would be a good manned position to control the drones. An SR-71(but not that aircraft specifically) telling a cloud of drones below it what to do also benefits from being nearly impossible to shoot down, with a reactive drone force under it to deal with incoming problems (as they will have plenty of time to spot a threat even if it is hypersonic).
There NEED to be AI agreements like we have for biological and nuclear weapons as well as genetic research (no cloning humans) otherwise there will eventually be a race to trust AI with broader and broader ranges of outcomes as the AI might come up with solutions we can't think of. And that's where both new capacities and new risks really start to emerge. Just like screwing around with genetically modified anthrax, that's not a good place to be. We've largely done that internationally in many domains (including China cracking down on human cloning) so it's not impossible to do.
I agree with what you have to say, but I think a big difference between AI and CBRN weapons is that AI has such a potential to improve civilian life as well. By limiting weapons development, we could also be stifling innovation. I guess it comes down to a matter of opinion at that point.
@@lukevillarreal1458And so far there are not major distinctions in the structure between warfare AI and civilian AI. In nukes for example you can track uranium mining, you need a lot more starting material to extract the correct isotope. Monitor the choke points and you control access. Idk how that would be done with data. The NSA could not stop cyber weapons (WannaCry ransomware) from getting stolen by N. Korea and used against the world.
@@lukevillarreal1458 Well, so does human cloning and embryonic genetic manipulation. Maybe even more so. But countries have agreed to put a ban on that kind of research. The big difference is that AI is being developed by some of the richest corporations in the world and it's hard to get them to stop or even pause.
@@flyingsword135 Very few countries are making chemical weapons and none are making bio weapons that I've ever heard of. There are also strict rules against using those weapons. You might remember that was the official reason for the US to go and invade another country. And you can thank your pearly pink panties that no bio weapon was released in 2019! First, any sane person has to admit that the evidence is not concrete one way or the other for COVID's origins. We just don't know (thanks in no small part to China). Second, it was NOT a bio weapon! If that had been weaponized smallpox or anthrax, we wouldn't have lost millions, we would have lost billions of people. If it did come from a lab it was an enhanced viral study that went wrong which is about the difference between Chernobyl (civilian accident- very bad) versus a nuclear sub launching all of its nukes at major cities (military accident- catastrophically bad).
Yeah really good point about the deployed systems. They are only learning in a controlled env. They do not learn/self mutate while deployed. Then they get locked down and deployed in an immutable state, once you know their architecture will produce good output for a given input. In theory however, if the deployed network, were being tuned and trained and tested by, yet another network. It's possible that the external network, the network that is sitting in a lab somewhere, which is being used to optimize and train and tune the deployed network, could embed behavior triggers that weren't intended. But they'd be very limited, for the reasons stated in the video. The network can't update it's decision making in real time. Still, you could plant in sort of a trigger or a queue like "Miranda" and have the network suddenly open up hidden alternate output pathways for a given set of input. You can tell this guy knows his stuff. Great watch. Crazy to hear him talk about flying against one.
We used AI for the Army Automated Air Load Planning System for the air lift planning for tbr 82nd back in 1984/5. It was used for load testing the C-17.
Pako, this coming from person who has spent 40 years in the computer world, you explained this kind of AI exceptionally well! I am not in the AI world but I keep a breast of things going on in it. And what I am hearing that advancements in AI was 3-6 months but very soon it will be in weeks, if it's not already there. An interesting point you made also was that industry is ahead of the military in AI and that shouldn't be. Because our enemies have boots on the ground with eyes and hears and will steal that information and send it back to their home.
You say that as if there aren't plenty of spies in the military and MIC. There's also an industry for retired US generals and admirals to get very lucrative "consulting" deals worth millions or tens of millions from foreign countries.
I remember when I was growing up there was a 1983 movie called Deal of the Century with Chevy Chase,, who plays an arms dealer, and it featured an unmanned fighter plane called the "Peacemaker" made by a fictional company called "Luckup". That was 40 years ago but that fictional drone basically was shown to be able to do all this stuff the real ones are supposed to be doing now.
What you guys seem to be talking about is an enhanced OODA (Observe, Orient, Decide, and Act) Loop in AI format. It makes sense that the drone with the best and fastest OODA Loop will always be on offense and the enemy drone will always be reacting until until its OODA Loop can't keep up and looses. The other thing that strikes me is that the US needs vastly improved air defenses to deal with drone swarms and to prevent drone ISR from locating friendly positions and movements. The Ukraine experience seems applicable in that Russia has a good capability to conduct drone ISR followed with Lancet 3 strikes. It seems to me that it might help to improve the Trophy system to defeat incoming missiles (i.e., Javelin types) and incoming drones (i.e., Lancets). But that won't solve the problem for the smaller vehicles and towed artillery. While the Valkyrie has about a 3,000 mile range, I wonder if it can keep up with manned fighters and bombers. Lots to think about, but I also wonder if there is any development for anti-submarine (either air or subsurface) to assist in locating and neutralizing enemy submarines. Finally, is there any development being done for a stealth drone capable of carrying the larger long range antiship and land attack missiles
This is reminding me of the Tie Droids on the last level of Star Wars: Rogue Squadron (N64). Only way I could get those was to switch to scramjets, gain distance, and turn around. Otherwise they just stay, not on my back firing at me, but just settling for a “tie game”, and stalling me out while the world devastators suck up everything.
One advantage of a UCAV that didn't appear to be talked about was how different the airframe would be. You are talking about vehicles that could pull off maneuvers that would kill a human pilot. They also might save weight by cutting things like life support features, control panels, and all the safety features. In theory this should lead to a more lethal air frame with fewer design constraints.
Lets not forget it wouldnt only just save weight and would allow them to design something smaller or have longer duration and payload
That's a good point --- I can't imagine a human fighter pilot going up against drones that can pull 50 G's with ease!
Just make sure to put in ai override...just sayin
The distinctions between a plane and a missile will blur. The distinctions between a plane and a drone will blur.
The design differences between the F22/F35 and the current unmanned fighters are drastic visually. They are smaller, flatter, more stealthy. Some are tailless (similar to the B2 and B21). No cockpit, no ejection system, no canopy as visibility is unnecessary. No manual controls. No oxygen necessary. No titanium or better material bathtub.
Love when you have him on Mooch! It's always extremely informative and definitely a great show.
Thanks! ❤
Paco is one smart dude.
Incredible content! Paco was outstanding with clear and concise explanations. He has made major improvements himself as he described the complex evolution of electronic warfare and the force multipliers that can be brought upon current adversaries. Absolutely riveting. Kudos to both of you for this presentation.
Pako gets the cigar for his articulation of applied machine learning in this domain. A+.
It made me chuckle to hear Pako say that he wanted a Moochie this year because as I was watching through this video, the one recurring thought I had was "this is one of the best interviews I've ever seen on this channel". Extremely informative. And I say that as someone who normally thinks that Bronk's interviews are just superior to everything else. Not only was this extremely informative...and on a very interesting, very important topic...but I'd say that Pako has gotten more comfortable doing these types of video interviews as well. This one seemed very natural. This was just a truly great interview all around.
This is a journey for all of us, and I'm just happy to be along for the ride.
A superb interview on an emerging hot topic. Clear, articulate and highly relevant. Thanks, Ward.
Thanks for the kind words!
The scary thing isn’t about our AI “checks & balances”
- it’s the other guy’s “checks & balances” that worry me!
Our government doesn't have any real checks & balances either
FUBAR is the universal balance. @@flyingsword135
@@flyingsword135 The entire U.S. government was originally set up with checks and balances and still has them, notice the present imbroglio.
With AI, w/o human in the loop, it's not checks and balances (there aren't any after deployment), it's do we understand the risks. I'm doubtful we do. But I think we'll probably deploy until the AI accidentally takes down an air liner and reassess.
@@PatrickProctor-Brown-xf6nd my prediction is that enemy shoots on purpose a massive missile into opponent's big oversea base, claims it was AI that shot accidentally to evade responsibility. but you cant see the insides because its military top secret. only their modified report.
Your knowledge and great floor plan of questions and good friends makes each episode a great event.
Looking forward
That rocket-assisted Valkyrie launch clearly reminded me of a BOMARC (1952-1972) launch with the exception that BOMARC whas vertical when launched.
I think they could carry a nuke warhead too.
Yea, as far as we know nuke-tipped BOMARCs are the only nuclear weapons Canada ever based on our soil.
Nuclear SAMs lol. I shudder to think what would happen if we tried that today, with society so reliant on personal electronics.
I thought of a Snark.
You forget the Genie missile on the CF101s
Paco deserves a moochie!!! The guy always brings it, and I learn so much from his interviews.
I'll have to find out what a Moochie is.. 😁
And that all the fighters stuck on the Ground are stuffed
I have been wanting to leave a comment for some time, but for the most part I watch videos on the TV so I don't have the opportunity to comment. I truly enjoy your videos and find you to be a VERY reliable and authentic source of information. I REALLY enjoy the episode where you have Paco on, he is fantastic! Keep up the great work! And thank you!!
As a retired researcher/teacher of AI, I want to say that it was a joy to watch Pako, simply because he was spot-on with every single thing he said. (Almost 30 years ago, I worked with Mica Endsley and a grad student on training artificial neural networks to emulate the decisions of expert pilots in air combat simulations. Mica, a major contributor to situation-awareness assessment, went on to serve as Chief Scientist of the Air Force.) I am deeply distressed by the volume of BS about AI I see in the mainstream media, especially at the New York Times. It genuinely brought a big grin to my face, seeing Pako nail the particular topic, and also the general state of affairs in AI, as he did.
Thanks! ❤
Indeed, I can no longer hang out with my buddies when they head out for major beer consumption. The talk turns to speculation about where all this Ayy-Eye might be leading us, reeling off a load of B-movie cliches that were bad science fiction 50 years ago. I end up comforting them that evidently, saturation beer bombing would exterminate any sign of superior intelligence within 50 km.
I mean if you don't know NYT is satire by now...
*It takes 8 hours approximately for a human to read 100 pages of book, it takes days and weeks for AI to read the same because both have profoundly different way of processing and learning.*
@Only_God_Is_Allah_SWT Not recently but not long ago Public Broadcasting Service was interviewing folks who said, "We are now getting answers to questions we hadn't thought to ask. So. Maybe AI must ponder character when reading some kinds of work.
I can't be the only person on here who sleeps slightly better knowing fine gents like these are on my team. Two outstanding humans.
❤❤❤ thanks for the kind comments.
Nope. You're not.
I'm with you 100%.
I agree about these guys skill level, but I can’t sleep easy knowing that the Matrix is no more than a couple decades away…….
@@bcdrummer1962 Everything can kill you, why worry?
@@tedmoss❤❤❤❤❤❤❤❤❤ ouch !
Garmin now makes a general aviation autopilot system should the pilot become incapacitated. There is a button on the center cockpit panel that the right seat passenger can push to activate. The airplane will then go into emergency mode and land at the nearest feasible airport. The system takes into consideration the aircraft's position, fuel, winds aloft and nearest available useful runway and prepares to land. The radio will transmit an emergency transmission on 121.5 mh and the transponder will squawk 7700. The intercom will instruct the passenger(s) what is happening and what to expect. It can do this in IFR conditions too. Garmin even advertises this in popular aviation magazines.
What got AI to do with this?
@@TimToussaint If civilian equipment can do this. Imagine what defense systems are capable of? AI is explained at the beginning.
I'm sure they'll have downloaded offline GPS maps that are new, plus terrain following, not to mention the DoD has an emergency satellite they can launch that'll work for 24 hours, I think? Maybe it was 48, I'm sure when I read that it's gotten better.
But that's pretty dammed cool regardless, I wonder if that last plane crash that was heading to DC into restricted air space where people heard the sonic boom a month or two ago where the fighters were trying to figure out wtf was going on and got in front to deploy flares to see if they were alive or awake cause it did do a u-turn and went back towards too what appeared where it took off from but crashed, maybe it ran out of fuel since it wasn't supposed to go very far, it was a weird scenario regardless. Mooch did talk and show on a flight record of it, then again I suppose if it had that program in it, it would've just landed in New York where it was originally going, I think? Hell, I dunno that one was freaky, plus it was the airplane CEOs family as well? Really freaky. Must've lost cabin pressure for both pilots to be knocked out.
I wouldn't be surprised if the military haven't talked to them about this or already have their own backup GPS system in space that's for their own use if something like this comes about.
It's pretty cool technology regardless.
People fly airplanes. Pilots fly helicopters.
@@pkelly3463 I flew airplanes. My good buddy flies Bell Jet Rangers. There is a big difference.
Wow! Great interview. Amazing stuff.
Much appreciated!
Great share. There is so much talk about AI right now. Being almost 65 and watching some of the movies we watched as kids how can I not be scared of AI? “Do you want to play a game” sticks deeply in my mind. Thank you again for sharing.
AI represents potentially great power. And power can be used in helpful or harmful ways. So maybe rather than being scared a responsible response is to be very cautious. Which the military seems to be doing, but how long that will last is another.
@@davidsmith8997 how can you be cautious instead of scared when you/I have no control over any of what will be done with it? As well as different militaries all over the world utilizing it with no possible way for you/I to know what they are doing with it?
AI is very much like the atom bomb race in the 1940's --- it's a necessary evil because you know your enemies are working hard on it as well
I'm right with you age-wise, and when I was "War Games" age I had a girlfriend named Jennifer...
At 80, I have thought about AI for 65 years now, and I am not scared of it, in fact, I use it in my Tesla with FSD. (The wife is a little scared, but she is learning). A great time to be alive. I am a Vietnam War vet and amateur radio operator, WA7VQR with 50 years computer experience. Edit; I think my wife is more scared of my driving than the car's driving.
Here's my vote for Pako's "Moochie", and my ongoing thanks for again presenting such an engaging, important and rapidly changing topic in the now familiar and welcome WC "dueling banjos" (or is that guitars?) format. The two of you work so well together, and contribute something that always adds up greater than the sum of the parts, and any government body seeking a similar update would do well to consider bringing you guys in pronto. Just as a video with two authors (think Shattered Sword) provides an amazing and captivating dual perspective, each of you has invaluable experience to bring to such a gathering of government officials . . . and one that we hope you will be bringing back on home here again soon.
Thanks for the support!
Ward, this was a really great video.
Always great to see Pako on here! Thanks for another great one, Mooch.
❤❤❤❤
Hey Mooch love the Marc Bolan Tee.
Paco is the man!!! Thanks Mooch!
❤❤❤
This episode should be nominated for a Streamy Award under the Subject Award category. One of _the_ best and most interesting discussions where you don't need a ton of visual clues to understand and be engaged with the dynamics of the subject matter. My favorite Ward Carroll interview to-date, alongside with Kim "KC" Campbell's episode.
Wow, that's putting us in some exclusive company! Thanks ❤❤
I love Pako's ' The Merge'...great info across a spectrum of topics I follow. Also excellent discussion on the varieties of AI...which is mostly lost in the media and in the public domain.
Thanks, I appreciate it! 🎉
Fantastic Episode, It has been a while since I have watched the channel. I glad I cam across it again.
Welcome back!
Love the Tshirt Ward. It’s interesting how AI changes risk profiles of missions and makes me wonder what super risky missions would entail (is this an escalating self reinforcing process?) as well as how adversary AIs will respond to each other in massive swarms.
Great topic and interview by Ward. Marc Bolan T shirt? T.Rex takes me back if that is him.
I think wars fought 70 years from now will basically be video games --- you'll hear a blurb on TV about "today U.S. drones shot down 50 Russian drones" and no human lives lost --- wars will actually be FUN but expensive LOL
Gracie Slick?♈️
@@fredwerza3478 there’s an episode of the original Star Trek like this except the losing side has to exterminate a set number of its own citizens at the end of the computer war game.
Upon your suggestion, I watched Unknown, Killer Robots. A good piece that helps you to to see the ying & yang of it all. That said, the brilliance of the technology staggers me! Thanks for bringing Mike on the show.
Happy to be here! ❤
Great conversation. Paco was really able to define the components and the roadmap of this technology, clearly and succinctly. Perception -> Cognition -> Action being the phases. Also a well said explanation of AI and machine learning which is so often clouded in buzz words by the public. Complex processes which are deterministic and happen the same each time, by instructions are not AI. But heuristic decision making, where choices are made with probabilities and based on learning from countless past data points is Machine learning.
Fascinating banter with two incredible pilots! Great to be a grunt fly on the wall and listen to a whole new world of war! And the comments from equally qualified people are great! Thank you!
Brilliant as always! Commander, you are the best!!!
Wonderful interview. Such a welcome addition to the channel with elegant, linear, articulate communication skills.
Thank you for the kind comment.
That documentary is very informative. And surprising.
What an outstanding video, Mooch. Please have Pako on as much as possible. His articulation of issues is BRAVO ZULU.
❤❤❤ Thanks for the support!
Well, plus they can pull as many g's as the airframe can handle, you save weight since no life support or even just seating are needed, plus the ship can be much smaller since you don't have to fit a man or men inside. Hey Ward, remember the HiMAT test drone?
Well it's gonna use those spaces for fuel
Nobody cares about pulling Gs with threat avoidance/attrition at distance in the networked kill web.
@@Carlos27thFS
Think you need to rethink or just initially THINK that comment. Let's just help you out by breaking it down, shall we, like sounding out unfamiliar words in kindergarten.
Trad combat platforms have both pilots and fuel ..... yes.
Fuel is usually but not always in the wings.
UCAV needs fuel but can completely dispense with the volume and mass of pilot LS and control systems. How does this equate to "gonna use those spaces for fuel". There is no need to use those spaces for fuel. Thus, those craft gain a significant performance advantage, in both thrust to weight and max G limits. If there are aspects we are missing, please explain in detail.
You take the pilot out of the plane and it doesn't have to 😩eat, sleep, or shizola 💩, so you have a force multiplier.
@@jamesburns2232 And young female warfighters have this shocking habit of getting knocked up! And then having to take extended leave or even full separation form service. Who could have expected such a thing to happen? After all there are REGS to stop that sort of thing from happening! (Tongue firmly in cheek).
Paco is an amazing human being! Thanks for bringing him on.
X-47, XQ-58 Valkyrie, MQ-25 Stingray and the AI F-16s are providing tons of valuable data. Can't wait for the Loyal wingman & Ghost bat to go into service.
Ghost Bat go in to service 2024 and is providing heaps of data actually
Incredible. Would love more episodes on this topic.
One of my favorite guests on your show...Mark is a cool dude
Who is this Mark cat? 😂
@@TheMergeMedia my bad..it was a typo...just subscribed to The Merge.
Great Episode! keep em coming.
Thanks!! ❤❤
As an ace combat veteran, i know how difficult it is to go up against ai combatants.
They can pull maneuvers far beyond what is capable for any human, the only way to defeat them in air combat is to be the main character.
May god have mercy on our souls.
The future of combat will end up AI vs AI. Then AI will determine man is inferior. Those Terminator movies no longer seem so far fetched now do they?
@@thystaff742whoa, hot take lol
@@thystaff742Lol they wouldn’t be allowed to do anything to it’s homeland people… That’s silly. We still have yet to have actual “ai” this last 5 generations don’t even know what Artificial Intelligence actually even is, we aren’t even close to getting there. All we can do is program mls. That’s IT!
The weapons are still junk as compared to the “code”. It’s best to know that our adversaries aren’t sitting on their hands.
So, have you found a reason to fight yet?
Very informative video. Thanks much.
"Get him close, we'll slam on the brakes, he'll fly right by us" - Maverick
😂😂 so true
I'd like to go on record saying that I'm all for the creation of Roko's Basilisk. You've got no beef from me, big guy
It never occurred to me that the drone have situational awareness of the complete battle and probably know all the specs of its opponent too, and can make decisions based on all those data-points every few milliseconds.
And it doesn’t black out.
The human brain can only process so much information per second --- a drone can have many brains and thousands of sensors --- a superior battle machine
But it doesnt have full situational awareness. That is hardware based. Does it have passive/active radar that is 360 degrees around, along with dual mode thermal imaging and uv imaging? Does it also have low light passive imaging 360 degrees? Has it been updated by all the supporting services as to all threats?
If the answer is "NO", then there is no such thing as "full situational awareness".
The F35 almost does, but even so, there is a ton of work to determine what to do, where, and when.
Using a UCAV in conjunction with a pilot in say an F35, that would be about as good as it gets, but it isnt impervious to various forms of jamming or interference.
Well at least it has occurred to you, so far it hasn’t occurred to the US Air Force and the US Navy, because those idiots are still planning on designing airplanes with pilots in them for the foreseeable future. What a joke!
@@watcher63034; World War II ended something like 78 years ago, that’s how far behind reality you are!!
Welcome from Pensacola
We can already see that the ratio -- of (A) hand-wringing about AI to (B) the unintended harm actually caused by AI in the real world -- is so huge that it borders on infinity.
People naturally fear the unknown. I believe automation in general has a higher error rate than AI does.
Until it actually backfires and kills people accidentally once. Then mankind will assign infinite value to whatever harm was done. Small probability, but intense measure if/when it does happen.
@@texasbeast239 that's why autonomous cars are taking so long to hit the road --- one death caused by AI will overshadow the thousands of deaths caused by distracted humans
You mean like people killed in Tesla with the automated driver?
You do understand that the "AI" used, is the same thing as the Beoing 737 max series were using , right? AI is only a machine operating a program. Nothing more.
Some dumbass probably said this about globalization 50 years ago. Responding to a crisis after the foreseeable problem arises is called incompetence.
Great episode!
I remember a stars and stripes episode from 2010ish regarding a few pilots going toe to toe with AI counterparts in some simulations, and they were shocked by how good it was. Glad to see this finally came into reality
It will be a lot better when those AI drones operate as a swarm and can pull 20+ g's all day.
@@surelyyoujokemeinfailure7531 Ghost Bat can operate in a swarm as is the name for it.
Ghost Bat was named after a Australian animal that hunts it's prey in packs using their high sensors.
Ghost Bat drone can do the same and why is most advanced AI drone as Australia builds good AI with swarm drone capabilities.
Paco is the best. I wish y’all could have talked longer. He said he couldn’t comment outside of dogfighting, but I believe the AI for BVR air-to-air and all of the components of electronic/jamming/SAMs warfare is where the bread will be buttered. Superior (AI) BVR will allow air-to-air dominance from distance. And electronic/jamming/SAMs (takeout) warfare will allow penetration of enemy defenses and space.
The drones could also carry smaller self guiding drones over the target area and deploy them in such a way they would be difficult to both detect and counter against. They could in essence be programmed to attach to either vehicles of personnel at a distinct asset level.
That was great, although much of it over my head and all of it outside my experience. Both of these folks are awesome to listen to.
So science fiction isn't a prediction but like....did people even watch terminator? Or Battlestar Galactica?
Contested perception is a really interesting phrase. It feels like a rerun of all the problems with object classifiers and adversarial images.
Sensor autonomy is gaining ground at a rapid rate, it’s the “platform” autonomy that’s the tough nut to crack.
One threat we will have to contend with from foreign AI adversaries is that they may lack the constraints we put on ours (they may not care if they shoot down a few of their own pilots as long as they get enough of ours). That creates a potential hole in our defenses.
There's no need to dehumanize our enemies. If their AI is dangerous it'll be because of incompetence not callousness.
@@appa609It is serioysly important for some of us to believe that 'we' use force with incredible constraint and reluctance.
@@appa609It is serioysly important for some of us to believe that 'we' use force with incredible constraint and reluctance.
Yes, humane German train drivers took Jews to humane camps, where they were treated very humanely. As in all socialist states.
Really informative interview, thank you Mr Carroll
As soon as I saw this I thought of the A.I. battle bots. The battles are over in seconds, and they are getting faster and faster. Also it scares me because, law enforcement moving towards A.I. might be the worst thing I can think of.
Unfortunately I think law enforcement is "all in" on AI --- it's gonna be too easy to use deadly force on people and then have local governments claim legal immunity from it because it wasn't a human pulling the trigger
AI Judge Dredd
This was a super interesting video. Very cool stuff!
There is absolutely no reason for AI to keep humanity around. That is the heart of the problem. AI will only keep us around until it figures out how to duplicate our ability to truly have a consciousness and useful creative thoughts. After that, we are done.
Science fiction. All AI is is mountains of probability calculations. That’s all it’s doing. Taking an input and finding the best answer based on statistical calculation. It doesnt feel, desire, or have consciousness.
Thank you for your service, Colin Hanks.
That's a first for me!
@@TheMergeMediaBetter then a Chet Hanks comparison. 😂
Times are changing. Technology is evolving, evolve with it or stay behind. And YES it will be hard. Lol
Thanks!
Fascinating interview Mooch. Your question at the beginning, "Will we have pilots in airplanes in the future" wasn't addressed. F-15E Strike Eagle Weapons System Officer Mike "Pako" Benitez provided some amazing unclassified illustrations and is unable to address Darpa classified issues. Appreciate your and Mike "Pako" Benitez's sides of the AI story.
❤❤❤ AI pilots are on the way. WHEN they will hit the scene has too many variables to predict. Only part of that is technology.
.@@TheMergeMedia Ghost Bat hit the scene 2 years ago and is in full service with RAAF in 2024.. you a bit behind in this technology lol.
Yes I'm receiving The Merge.
They will probably end up taking missions with less risk reward than you would only normally take if you were losing. So data will show "wow, we would have lost X number of pilots." And that will be a thing by manufacturers and their supporting high rank officers. What we will never see are the billions and billions wasted on high risk missions that should have never been taken because there was no human life at risk, but chance off total loss was high compared to mission goal or reward.
the man the myth the legend! great to see pako again!
❤❤❤❤
Pako speaks the truth. My mil days are long behind me but my current career in software is deep in AI. Machine Learning is the major thinking component of AI and his description is spot on. It learns through repetitive events, and we sim thousands of events per day to 'teach' the AI how to analyze and react.
The trick will be having enough variability of events to give the AI broad enough training to deal appropriately with all eventualities.
SEAD seems like an area for early opportunities for success with AI. Instead of 'Wild Weasel' when humans are involved that will be more like MILD Weasel from a human safety standpoint.
Thanks!!❤
Mike is one of the smartest people you've had on. He's incredible!
Very good discussion. You're probably right... highest automation is expected. However, as we remove humans from danger, the human's natural desire to avoid the fight (start the war) may to down. It may lead to more war, not fewer deaths. Secondly, how do you cause the aggressor to capitulate if he is only losing "things" not lives.
Wars have always been won or lost based on money spent --- using drones only amplifies that tenet
Money spent is a major aspect to winning, but the perception of lives lost is the biggest deterrent to wars starting, at least in the West.
Thank you for this outstanding content! Best I've seen lately.
It would be interesting to see how useful AI is when its sensors are degraded in a realistic combat scenario, jamming, direct energy weapons burning optical sensors, cyber warfare like introduction of spoofed gps data etc, supposedly humans traditionally have the advantage in this environment as they can sort the chaff from the wheat, wonder how robust these agents are when faced with that.
I wonder if laser communication would be helpful in bypassing EW in certain conditions.
Jamming should be a part of the past. Ultrawideband technology is still evolving - I was alerted to it by one of our communications engineers more than 20 years ago and the FCC is still catching up. The FCC is only concerned with civilian uses in the US, and only in regulated bandwidth. UWB in its advanced forms is virtually impossible to intercept or jam unless the "carrier" pattern of pulses is leaked. The bandwidth of such services is almost literally 3 GHz to daylight and receivers don't care about anything that does not match the "carrier" pattern. Better yet, the extremely low average power requirements mean almost unlimited service years or human lifetimes - without draining the power source.
AI can discern the race of a skeleton with 98% efficiency. Humans can't tell at all or know how AI can tell.
What makes you think a human would be any better under those same kind of impairments?
@@flagmichael Only in bursts is it nearly impossible to detect. Dont forget, every time you intercept a Russian Elint aircraft, that is literally what they are doing. Watching, listening, learning. Every time you communicate or pass messages.
It doesnt need to jam every frequency. It only needs to know which ones you use a lot. The whole idea of Growler type aircraft is either to jam specific frequencies, or to share jamming of all possible frequencies.
The first time a UCAV shoots down a manned combat aircraft, it will set a tone for this new arms race.
It bypasses some barriers to technology-sharing because we guard certain manned combat technologies from FMS.
Imagine UCAV operating in Ukraine right now, for example, especially with a cooperative OCA/strike mission set.
Well if you count the weaponized toy type ucav used in the Ukraine, they have already destroyed manned helicopters with them.🤔
It was attempted before when an MQ-1 Predator drone that had a stinger launcher loaded tried to take a pot shot at an Iraqi MiG-25. The Foxbat fired an R-40 in response and the R-40 had a large enough exhaust that it distracted the Stinger causing it to miss. The R-40 missile then splashed the MQ-1 Predator. First ever engagement between a UCAV and a manned fighter turned out to be a defeat for both the USAF and UCAVs. That said an MQ-1 was probably not the best choice of UCAVs to use against a Foxbat, likewise a Stinger not the best weapon to use either.
As for Ukraine I have seen video of a drone engaging another drone. One of their FPV operated drones with a bomb attached to it crashed into a quadcopter the Russians were using for reconnaissance.
@@NuclearFalcon146 MQ-1 isn’t a UCAV though. UCAV is autonomous, jet-propelled, LO/VLO, with a heavier payload and higher speeds. When a UCAV launches an AIM-120C7 through D3 at another fighter from unobserved parameters and splashes that fighter, it’s going to be a game-changer. Same for a transport plane or AWACS, which are even bigger trophies.
It will also be a big deal when one or more of them conduct a deeper penetration strike mission with JDAMs.
@@LRRPFco52 UCAV = Unmanned Combat Air Vehicle. It includes remote operated and autonomous. MQ-1s, MQ-9s, and even the Ukrainian quadcopters rigged to drop grenades count.
@@NuclearFalcon146 I'm referring to the UCAV program constantly whispered about dating back to when we were on Global Hawk. UAS is probably the term you're looking for.
Great talk! The explanation of reinforcement learning was great!
There’s a lot of awesome AI stuff going on in gaming. Theres an AI project for a team based game that is quite complex. The AI players can work collaboratively with human players, and keep situational awareness of the whole map while in an intense frenetic competition. Something that is incredible to watch, and no player even the worlds best can match. I think one of the big uplifts will be that not only are the airframes simple and relatively cheap, but more airframes in the air at once is a multiplier. With newer radar etc there is a network effect, in addition to the redundancy and capacity of higher airframe numbers.
We should embrace these technologies like these because honestly, we have no choice, they are coming and whoever doesnt have it will be the little brother at the dinner table.
Scarcely a decade ago I would have hesitated to agree. Today I have no such reservations.
Same rationale was used to proceed with atomic weapons development back in the 1940’s.
@@glennac Right. And as I've said elsewhere, we've agreed to limit chemical and biological weapons. We've agreed to not research human cloning and some types of genetic research. Seems like the sane thing to do would be to agree to something similar with AI. Anyone caught breaking the rules would be treated like someone caught developing weaponized smallpox or genetically engineering soldiers.
Short story called Revolt of the brains by Stephen Marlowe writing as CH Thames imagination stories of Science and fantasy December 1956
What if China camouflages their drones with fuzzy letters and pictures of crosswalks? Will a 6th gen. WSO be necessary to help the AI pass all the captcha tests?
If it has a captcha, just shooooot!
seems like having an ultra-high ceiling, insane speed control craft would be a good manned position to control the drones. An SR-71(but not that aircraft specifically) telling a cloud of drones below it what to do also benefits from being nearly impossible to shoot down, with a reactive drone force under it to deal with incoming problems (as they will have plenty of time to spot a threat even if it is hypersonic).
There NEED to be AI agreements like we have for biological and nuclear weapons as well as genetic research (no cloning humans) otherwise there will eventually be a race to trust AI with broader and broader ranges of outcomes as the AI might come up with solutions we can't think of. And that's where both new capacities and new risks really start to emerge. Just like screwing around with genetically modified anthrax, that's not a good place to be. We've largely done that internationally in many domains (including China cracking down on human cloning) so it's not impossible to do.
I agree with what you have to say, but I think a big difference between AI and CBRN weapons is that AI has such a potential to improve civilian life as well. By limiting weapons development, we could also be stifling innovation. I guess it comes down to a matter of opinion at that point.
@@lukevillarreal1458And so far there are not major distinctions in the structure between warfare AI and civilian AI. In nukes for example you can track uranium mining, you need a lot more starting material to extract the correct isotope. Monitor the choke points and you control access. Idk how that would be done with data. The NSA could not stop cyber weapons (WannaCry ransomware) from getting stolen by N. Korea and used against the world.
Umm countries are still making chem and bio weapons. A bio weapon got out of a lab back in 2019...
@@lukevillarreal1458 Well, so does human cloning and embryonic genetic manipulation. Maybe even more so. But countries have agreed to put a ban on that kind of research. The big difference is that AI is being developed by some of the richest corporations in the world and it's hard to get them to stop or even pause.
@@flyingsword135 Very few countries are making chemical weapons and none are making bio weapons that I've ever heard of. There are also strict rules against using those weapons. You might remember that was the official reason for the US to go and invade another country.
And you can thank your pearly pink panties that no bio weapon was released in 2019! First, any sane person has to admit that the evidence is not concrete one way or the other for COVID's origins. We just don't know (thanks in no small part to China). Second, it was NOT a bio weapon! If that had been weaponized smallpox or anthrax, we wouldn't have lost millions, we would have lost billions of people. If it did come from a lab it was an enhanced viral study that went wrong which is about the difference between Chernobyl (civilian accident- very bad) versus a nuclear sub launching all of its nukes at major cities (military accident- catastrophically bad).
Yeah really good point about the deployed systems. They are only learning in a controlled env. They do not learn/self mutate while deployed. Then they get locked down and deployed in an immutable state, once you know their architecture will produce good output for a given input.
In theory however, if the deployed network, were being tuned and trained and tested by, yet another network. It's possible that the external network, the network that is sitting in a lab somewhere, which is being used to optimize and train and tune the deployed network, could embed behavior triggers that weren't intended. But they'd be very limited, for the reasons stated in the video. The network can't update it's decision making in real time. Still, you could plant in sort of a trigger or a queue like "Miranda" and have the network suddenly open up hidden alternate output pathways for a given set of input.
You can tell this guy knows his stuff. Great watch.
Crazy to hear him talk about flying against one.
Love the T-Rex shirt, Ward! 😀
We used AI for the Army Automated Air Load Planning System for the air lift planning for tbr 82nd back in 1984/5. It was used for load testing the C-17.
Looks like everyone is getting so excited about building that Skynet thing. Just like in Terminator...
Thought provoking...to say the least! Thanks Pako and Mooch!
Pako, this coming from person who has spent 40 years in the computer world, you explained this kind of AI exceptionally well! I am not in the AI world but I keep a breast of things going on in it. And what I am hearing that advancements in AI was 3-6 months but very soon it will be in weeks, if it's not already there. An interesting point you made also was that industry is ahead of the military in AI and that shouldn't be. Because our enemies have boots on the ground with eyes and hears and will steal that information and send it back to their home.
You say that as if there aren't plenty of spies in the military and MIC. There's also an industry for retired US generals and admirals to get very lucrative "consulting" deals worth millions or tens of millions from foreign countries.
I remember when I was growing up there was a 1983 movie called Deal of the Century with Chevy Chase,, who plays an arms dealer, and it featured an unmanned fighter plane called the "Peacemaker" made by a fictional company called "Luckup". That was 40 years ago but that fictional drone basically was shown to be able to do all this stuff the real ones are supposed to be doing now.
Lol your little Dr. Strangelove reference was great "Mr. President, we cannot allow a mineshaft gap!"
An excellent and also horrifying discussion. Thanks !
What you guys seem to be talking about is an enhanced OODA (Observe, Orient, Decide, and Act) Loop in AI format. It makes sense that the drone with the best and fastest OODA Loop will always be on offense and the enemy drone will always be reacting until until its OODA Loop can't keep up and looses.
The other thing that strikes me is that the US needs vastly improved air defenses to deal with drone swarms and to prevent drone ISR from locating friendly positions and movements. The Ukraine experience seems applicable in that Russia has a good capability to conduct drone ISR followed with Lancet 3 strikes. It seems to me that it might help to improve the Trophy system to defeat incoming missiles (i.e., Javelin types) and incoming drones (i.e., Lancets). But that won't solve the problem for the smaller vehicles and towed artillery.
While the Valkyrie has about a 3,000 mile range, I wonder if it can keep up with manned fighters and bombers.
Lots to think about, but I also wonder if there is any development for anti-submarine (either air or subsurface) to assist in locating and neutralizing enemy submarines.
Finally, is there any development being done for a stealth drone capable of carrying the larger long range antiship and land attack missiles
Boyd's legacy indeed. Thanks for reminding everyone about OODA loops - it's shocking how many people don't understand this concept.
Well well well...this will be a hard fight for the moochie award. Great stuff from Mike.
Skynet is coming. Craziness. Hearing about this tech definitely brings up a lot of questions.
Paco is a great lecturer.
You should show us your plaque, trophy, patch, memorabilia collection from your time. Would be neat to see all that stuff
Nice Force multiplier AI drone wingman
Great informative discussion.
Good Video as usual
This is reminding me of the Tie Droids on the last level of Star Wars: Rogue Squadron (N64).
Only way I could get those was to switch to scramjets, gain distance, and turn around. Otherwise they just stay, not on my back firing at me, but just settling for a “tie game”, and stalling me out while the world devastators suck up everything.
Reminds me of what Chohei Kambayashi depicted in his science fiction novel YUKIKAZE...which was first published in 1984.😮
Will we get Variable fighter’s and pop idols as well I wonder 😉
Great interview
No body can stop it" good thing the good guys are on it!
Amazing.
Pako is most deserving of a Moochie