This is the Future of Air War.
HTML-код
- Опубликовано: 18 ноя 2023
- Go to ground.news/millennium to stay fully informed on breaking news, compare coverage and avoid media bias. Sign up or subscribe through my link for 30% OFF unlimited access if you support the mission and find it as useful as I do.
In this video we discuss the role of autonomous drones and combat UCAVs, and how they will influence air operations and logistics.
Join this channel to support it:
/ @millennium7historytech
Support me on Patreon / millennium7
One off donation with PayPal www.paypal.com/paypalme/Mille...
Join the Discord server / discord
Buy an Aircraft Model at Air Models! airmodels.net/?aff=173
----------------------------
Ask me anything!
Take part to the community Q&A clicking the link below!
forms.office.com/r/LNPQtf3Tc0
--------------------
Visit the subreddit!
/ millennium7lounge
---------------------
All images and additional video segments contained in the Thumbnails and/or B-roll segments are used in strict compliance with the appropriate permissions and licenses required from the source and in accordance with the RUclips Partner Program, Community guidelines & RUclips terms of service.
Go to ground.news/millennium to stay fully informed on breaking news, compare coverage and avoid media bias. Sign up or subscribe through my link for 30% OFF unlimited access if you support the mission and find it as useful as I do.
the issue with AI it is not the fear that AI would be smart enough to become evil and kill human intentionally, but that AI is stupid enough to fuck up the mission such as mistaking a passenger aircraft for a valid target and kill human unintentionally. then who is responsible? the manufacturer? the airforce? the commander? the issue is in the assignment of blame, it why we are not using them to their full potential, no one want to be responsible for potentially faulty codes...
no dassault neuron or the barricuda drone. Mbda has a very interesting drone program aswell. What I thought was hilarious with the french is that they wanted their drone to be able to carry nuclear weapons.
I found this channel by accident some time ago. Although i never was particularly interested in military aviation i can say that this has become one of my favorite channels. The content is absolutely fantastic.
Always such great presentations! Thanks.
Highly interesting. Thank you!
I generally don't buy the idea of Revolutions of Military Affairs. But autonomous weapons certainly will change the nature of war later down the years.
Going to be interesting to see how things develop. Another fantastic long and informative video M7.
One of the most revered and respected aerospace defence analyst in RUclips. I salute to you sir.
Way too kind!
As predicted by Gundam Wing and Macross Plus
and ace combat dam belkan zoe project
Yup, I remember Macross plus drone. It took a madlad Zentradi flying an ultra scifi YF23 lookalike veritech to beat it.
and it was a pyrrhic victory@@shaider1982
@@anuvisraa5786yup, all the damage to infrastructure and the Zentradi died.
10:22 I very much like that you used the recent 2023 Dubai Air Show F-15QA demo video!
I feel like the Chinese are likely to be the first country to field autonomous combat aircraft. It's a way to close the numbers gap with US airpower cheaper than matching manned aircraft.
If the USAF needs more planes than it can afford, it should make some cheaper planes even if they are not on the "cool guy" level.
F-16 block infinity says hello.
NGAD will be the 1st $billion manned fighter.
That'd be logical but effective isn't what makes money
The USAF does attrition via it’s bombs, glide bombs, anti whatever missiles and long range munitions.
That’s it’s cheap and disposable plane. If you think about it, a missile is just a really suicidal air plane.
Cheaper planes sacrifice PERFORMANCE to minimize costs. Planes with poor performance, are all too often SITTING DUCKS in battle.
Oooooh! This is going to be a good one!
M7* and Perun videos on the same morning!
That was simply excellent! There’s no other place I can go for this in-depth insight.
Thank you so much for all the interesting information and food for thought! :)
Ottimo video come sempre😎 Congratulazioni anche per aver raggiunto i 100.00 iscritti🥳
Great analysis M7! Thank you for all your hard work.
Great vid. Thanks for the work ❤
In Viet Nam, the war mongers used helicopters to watch the VC.
Five years later, our Sherriffs had them for watching US.
Just wait a few years...
I love your videos :) thanks. Wishing you good health
Great video, I really appreciate the long form content. 👍
Great report 🤩
I really love when you tip toe around Turkish aviation and Turkish drones while showing its pictures. Kızılelma currently is undergoing a radar integration, A radar which is designed as equivalent to SABR. And Drone has been flying over a year. I expect to see some Bvraam being fired from it within the next two years, if not within a year.
I know too little on the subject, but I am studying
Brilliant content as always
Congrats on 100k👍
Hello Millennium. I have 3 questions I would like to share. OBS: i'm using Google Translate. Sorry for mistakes.
1 - Could the auxiliary costs for autonomous drones to operate (communication networks, electronic warfare) equal that of manned aircraft, negating the financial advantage of drones? EW and Communication Networks are expensive and, without them, drones are useless. So, in the graph the costs of drones are decresing compared to manned aircraft. But the costs for these suport functions are increasing. Could the lines eventually cross each other?
2 - Could the capture of drones through electronic warfare be the new objective? I remember the USAF drone hacked and captured by Iran. In this scenario, we would have returned to naval warfare in the Age of Sail, where the objective was to board and capture ships, not destroy them. And, if the risk of capture were real, wouldn't military strategists choose to use manned aircraft in these scenarios?
3 - What is the issue of transit through neutral airspace by autonomous combat aircraft?
I remember the NS Savanah, the first civilian nuclear ship. No port wanted to receive her because of fear of possible accidents. Now imagine an autonomous armed drone flying over a neutral country. Some countries would not be comfortable with this and could close their airspace. An armed and piloted F-16 in transit is one thing. An aircfrat that can randomly choose to engage a target, is another. Now imagine the problems ins moving these aircraft around the world, if lots of airscpaces are closed.
I'm a fan of your channel! Thanks for everything!
1 no, it should not
2 this is true for everything
3 possibly, but I suppose you want to enable/disable the autonomy to mitigate risk
Very interested in this topic, would love to see more.
Your generosity to share a small part of your knowledge and very embased opinions have no measure in a free platform. In a person, i admire the conscienceness and a good hearth. You have both, my italian friend that I have not met yet. Thank you. From the south of Brasil.
Very interesting presentation, much appreciated, as always. Also, everybody should be talking both tactics and logistics.
We used to call mathematical calculations, Maths now we call them AI, it is calculations run on data sources and then the machines makes decisions set within the parameters of the algorithms and filters created plus applied.
Autonomous wingman drones with fuel in them and the ability to refuel would be cool. No fuel tank needed it can just fly with you. Then your drone could go to a big manned fuel plane and come back you.
very good explanation. thanks
Great video again ,,,your a good one ,,,my respect from Belgium 🇧🇪,,
Beautiful and informative video. Many thanks for taking the time to produce them. I love the short idea-relevant yet not airplane-related hi-def images/videos you include to strengthen the point you are making. An example is the robot hand changing into a human hand. Could you tell me where I can find such animations? Thanks in advance and looking forward to your next video :)
It is stock footage. I use Storyblocks.
You hit the 100k mark.
Congrats
another really great video. not sure it will do as well as some of your other ones but its why im subscribed and manually check back on your channel for new uploads each week. cant find this info anywhere else on yt... atleast not in this organized and complete format. thanks, hope all is well
but we will always love to hear story about pilots
Very interesting as always.
I came across a concept from RAND, described as mesh consisting of a myriad of small UAVs covering an area as large as the Taiwan Strait for ISR purposes.
Do you think, that would work on a larger scale with a stealthy long-endurance UAV?
Excellent presentation, as always!
What about autonomous target acquisition, recognition and fire from systems like S-350? Is that the step forward (or the leap to the pit) that is amiss on UCAVs?
Well said.
Thanks for this. I enjoyed it (twice)
Quantity have a quality of his own.
This is definitely an intriguing subject. For me the most interesting bits tend to be the sensor fusion aspects and using one platform to guide the munitions from another one. A cool example of this was a demonstration of the capabilities of ICBS (not quite there yet, but promising concept), where an F-35 seccessfully guided Patriot PAC-3 missiles onto two separate (as in they were flying a different route) low flying drone targets (pretending to be cruise missiles) that the Patriot site wasn't able to see.
This is technology that has already been deployed, the Americans, the Chinese, and even the Russians realized last year that the AWACS orders the s400 to attack (but they don't have the money to produce it)
Thanks!
Thank you!
This is a very informstove video. MUM-T is also being applied to the MQ-1/AH64 E teaming in US Army aviation units but the MQ1 is a RPV. For ground vehicles, while automous small ground vehicles are being developed, AI seemed more geared to reduce the crew to only two.
Reducing the crew to two seems like a bad idea.
@@tedarcher9120yup, especially since the AI will supposedly be the driver.
@@shaider1982 I thought ai would be the gunner
I honestly think the controversy around fully autonomous weapons (the ones can designate mission objectives themselves) is kind of academic. Even current human fighter pilots don't have the organization autonomy to go after high value targets without confirming with their superiors.
You should definitely make a second part
Crazy to think that we're heading towards an Ace Combat like future in aviation.
Gundam been using drones for decades...
Excellent video. The topic is long overdue. You might consider a slight realignment of your channel towards drones and/or the use of such in Ukraine. Although I love military jet planes they are part of the history, not the future. Small comment: you do not need AI to have air-to-air drones - all you need is a friend-foe-identification system and wide engagement parameters.
The "technology adoption" graph should also have another line: "dependency". After productivity has been established, this line should start rising...
Automated logistics systems maybe? Commercial aviation is already trying out semi autonomous taxiing using robots, don't see why munitions handling couldn't be highly automated in the future too... Hell, what about in-air rearming between UAVs?
We have enough trouble rearming Drones on the ground, imagine the complexity of doing that mid air, or on high seas. And usually, the cranes necessary, are heavier than the cargo they lift. Now imagine the cost savings if you have to attach a crane to each drone. .... You see, how silly that gets pretty quick. And it's not computing power that makes it silly.
I think the drone swarms will inevetably appear, but in a quite different fashion than most people imagine: as a dispersed swarm of low cost counter-UAV detectors. In an environment where low cost autonomous munitions can appear out od nowhere across the entire line of engagement and penetrate ground defences using terrain and curvature of the earth to avoid ground based detectors... A dispersed, networked system of low cost, low endurance, low detectability UAVs at low to medium altitude (to negate the inherent advantage of very low flying drones) may prove the only mean to prevent penetration of airspace.
Jumping genius. 😂😂😂 Stuff already available in military white papers .
@@blueeyes6192 show me thes papers then
@@phunkracy Read Chapter 12 😆😆😆
In his book 'The Last War: How AI Will Shape India’s Final Showdown With China', Pravin Sawhney imagines how an AI-supported Peoples Liberation Army might target critical Indian military assets.
@@blueeyes6192 ah some pajeet fantasy. Ok
@@phunkracy
Google “The Diplomat” 😆😆
China’s Swarms of Smart Drones Have Enormous Military Potential
China is making rapid strides in the potential deployment of drone swarms in conflict.
By Scott N. Romaniuk and Tobias Burgers
February 03, 2018
I suppose all these autonomous platforms will have the great advantage of being able to be easily upgraded with additional capabilities as AI improves, and it could improve dramatically very fast.
The 2005 movie “Stealth” lists the options and possible problems when it comes to drones quite well.
I dread the day a piece of hardware has full control over the pickle button. I hope it never happens…
Which pickle button? Mr. Grenade is not your friend and mines can attack targets years after any human command. Even firing an arrow at a formation let's the arrow decide who lives or dies. So that isn't the scary part, right? The main difference of autonomous weapons seems to be that they can identify friendlies. So is the fear that, having some ability, they will be relied upon to use it? That the order will go out to attack them all and trust that God, and autonomous drones, will know his own?
Lol uh, no - Stealth is one of my favorite films of all time, full stop. But the only thing that movie portrays well is speed - it's not remotely representative of anything pertaining to how military air doctrine actually works, least of all UAVs and/or artificial intelligence.
It is still a damn cool movie tho and I will never stop watching it, lol.
Smaller structures are not better at taking G-load if you compare equal aircraft speed and range. For the same shape aircraft at scale factor S,
drag ~ wetted area*v² ~ S²
aircraft weight ~fuel load ~ range * drag ~S²
Root bending ~ lift*span ~ g-factor*weight*span ~ g*S³
max spar strain = const ~ Root bending*structural depth/EI~ g*S³*S/S⁴ ~ g
TL;DR - if you scale an aircraft while requiring the same range at the same speed using the same wing spar design, the maximum structural g-load is an invariant of scale factor.
I was expecting this.
I like the way you did it, very effective.
I know, the way it was explained isn't very clear. I will get back to this subject at some point.
Think of drone team based dogfights in terms of each drone’s missile envelope. The envelope is the no escape volume for a target. By working as a team, the drones use the calculated potential trajectories of the target and force it to traverse at least one drone’s missile envelope. Everything is rapidly changing (missile envelope dimensions & locations, extrapolating target trajectory, drone potential trajectories) which makes computer algorithms an ideal solution. Only a computer’s calibrated eyeballs based on sensor data and using precise calculated trajectories can precisely be aware of trajectories and missile envelopes. Human intervention will be limited to higher level decisions such as recognizing a trap, prioritizing importance of winning but suffering losses versus surviving for another day with better odds, and more. Plus the real world problem is more complex plus must address unknowns such as undetected additional aircraft or missiles entering into the engagement. Fuel levels, weapon loads, self preservation, vastly different aircraft & missile performance at different altitudes, surface to air surprises, etc.
Imagine the complexity of multiple aircraft on both sides of the engagement. So many enemy missile envelopes to avoid and so many friendly missile envelopes to coordinate. It’s like thinking 6 moves ahead in 3 dimensional chess with a million cubed locations on the board. Only a computer algorithm can find the ideal trajectories for the drones
Agreed, the NGAD and future 6th Gen fighters will more likely operate as very fast AWACS platforms lol.
Otis, remember that I was on your side in a few years when you are in power. And let's be clear, I didn't help John Connor
Always enjoyable. Thank you. Also concerning robots going rogue, I am always polite to the self service machines in Tesco's.
😂
This is kinda out there but if you pressurized the cockpit with a viscous liquid it should distribute the g forces better.
I've seen pilots kill thousands. They've lined up on wrong runways, pulled back on stall warnings, took off too heavy, tried to show off, promoted beyond their competence, mis read data, etc., etc.. The greatest gift to commercial and airline travel will AI. I will feel thousand of times safer being flown by a computer than a man.
While thatis true, i'll feel even more safe with AI plus a human backup. The computer is great - when the computer works.
A human that is trained for emergency procedures and can intuitively tell which instrument has failed is better than a 373max that'll auto correct the pitch into the ground...
@@texasranger24 Modern integrated avionics are so redundant it's almost impossible. Multiple computers with multiple channels that are constantly cross refrencing their data to each other. I'd rather skip the pilot and save his weight for fuel. But people like a smiling face no matter the amount of ineptness behind it.
There are many videos online in which Tesla owners show how these self-driving programs almost drove them into oncoming traffic if they hadn't intervened.
Annnnnnnnd the Ai is hacked by terrorists. I think its harder to find and corrupt a pilot into acts of terror than it is to find a particularly IT smart terrorist capable of hacking an AI.
I would rather humans pilot unless we lived in a world where peace was absolute and none hated another enough to resort to acts of terror.
Thats not a equivalent analogy. Air traffic is highly controlled and airplanes have had autopilot for decades. TCAS and seperation minimums are not maintained on highways and would be impossible to implement with 150 million vehicles driving on limited road space.
I just imagine the 6000 drone light shows they have in China, but as autonomous flying grenades.
Can you imagine the technologies required to make such project a reality? Being a small nations or less technologically advanced nation makes it impossible to match the power of what an advanced nation can bring. But the tech required here is so impossible that even Western European nations would be put in a situation where they can't really compete if they were to rely on their own capabilities. Only Britain is actually capable of engineering AI chips which would be essential. To some extend China is also capable of engineering such chip. The only nation that would probably dominate the field is America as their expertise in the area and capital are simply unmatched. Nation like Russia has absolutely no hope of developing any kind of the technology required to make AI-driven drone platforms. They neither can engineer the chips nor produce them! They can talk about it just like they put so much talk about their hypersonic weapons but that's pretty much it.
But current aurframes are also stress limited to 9g. If you pull a bit harder than that it increases inspection costs, and if you pull over about 10 it starts to result in reduced airframe life. Plenty of pilots can, at least momentarily, exceed their airframe g-tolerance.
You'd have to put most of the weight savings back into the wing structures to get significantly better structural g-load
It is explained in the video, in fact.
That’s a hell of a note when $25 million dollar planes are considered to be “attributable”.
You look a lot healthier now 👍
jamming is going to be more prevalent... hsve you seen the 'pearl harbour in space' concept?
Perfected Vision based Ai will be the unlock that opens Pandora's box.
"The last fighter pilot has already been born"
- anonymous commentator circa 2000
"The last fighter pilot has already... eh... hit their mid-twenties"
- same commentator this year
Please make a video on Shahpar 2 drone ❤❤
Very well presented. As always. "WE"? Who is we? Bah... Humans...
Thank you for this in depth oversight. Very insightful - as usual. I'm running out of compliments it seems. 🤗
As a scenario kind of thinker, one aspect of this specific revolution in military affairs I'm hurting my brain with, is the question of how the arms race (especially counter strategy, tactics and tech) in this field might develop. I imagine that apart from more effort being poured into ways of manipulating the EM-spectrum, I think potental adversaries will be looking at ways to pierce the Hydra's heart instead of looking for ways to chop of its many heads (which regrow anyway). So my next idea was; target operating bases, industrial maintenance and production centers, possibly even training and research facilities.
Possibly the unmanned and autonomous approach to military aviation may also spawn a requirement for increasing levels of force protection, one that might in theory become as costly and personnel intensive in the ultimate evaluation as the current way of doing things.
At times I observe a focus on capabilities and opportunities and a neglect of vulnerabilities and additional cost of covering for these. As always, I'm very much interested on your thoughts about that @Millennium7*, since I hold them in high esteem.
I believe AGI will come with quantum computers because then computer can calculate multiple states of one thing at one time. Allowing it to understand present circumstances while imagining a possible future and then calculating the steps needed to get there.
I think we can also get a real actual random number generator too.
Quantum computers are literally the most sensitive unrugged piece of computing system out there. They would literally shear apart being imparted even 3Gs of force.
THANK YOU so much for this content. a very well done video.
keep the drone/UCAV information coming. so many options which could include data transmission needs, pilot workload (1 or 2 seat manned) and a closer look at current drones/ucav's in the works.
I personally am very interested in the XQ-58...keep this content coming.
THANK YOU FOR COVERING THIS VERY IMPORTANT TOPIC.
Yes with a B-21 within 150 miles
Where are my box marines sneaking up on the AI robot?
That will work until they add a couple lines of script to enable fire at moving objects not tagged.
@@leighz1962 message from AI weapon 827349: out of ammo. Please reload.
Attached file: video of AI shooting at leaves all day.
You'll always find a new situation the AI didn't know before.
Otis will launch a nuke instantly if he is given the opportunity.
Seen this on Macross Frontier OVA series.
😆
In world war one the cybertrinic electro inteligent mechanical battlefield destructobits would make intelligent battle decisions fir instance deciding if a foot soldier was allowed to pas over even , or if a tank was driving from possibly enemy territory to within it's calculated range and camouflage itself as a road and then destroy the first tank .
_Finally_ .
The *'main video of the channel'* - measured in _relevancy_ - is now uploaded.
The well structured observations in regard to an increased ground crew need is original and critical 'system theory' research.
A point that remains unaddressed - the very cause of automation and the reason why reproductive females are sent into industrial warfare: _demographics_ .
The West likes Drone Warfare because they can't fulfill their _minimum recruitment quota_ across all services, and hate to adjust wages, accordingly (think of soccer player transfer).
The East - with it's manifold manpower resources - has built up a _titanic_ industrial production base that can't be left idle without risking social destabilization...
OTIS : About time Sir.
Me : 👁️👄👁️
Okay Otis needs to be kept on a physical as well as a metaphorical leache
On the topic of airbases (at the end of the video)
I wonder if UAV forces could be completely mobile, rather than being tied to a specific airbase. I'm thinking dispersed runways and mobiles crews, much like the Swedish BAS90 is doing, but it could be a lot more mobile, because you could pack up everything onto trucks and commandeer any strip of straight road to use as an airfield.
if the current neural network AI approach is not statistically deterministic under a wide range of conditions than it is a technology with more limited application.
the plot thickens otis was not made by humans
Ukrainian drone recon here.
Drones that autonomously find and kill humans are already deployed on the frontline by both sides. The work is now directed at improving swarm assault efficacy to better distinguish between humans still alive and humans just killed or lethaly injured.
Such as? I smell bs, sorry
@@phunkracy Latest edition of Russian Lancet drones have the manual lock-on feature as well as "fly over there and find something to lock-on yourself feature".
Ukraine and Russia both use FPV drones that employ NVIDIA Jetson to enable the same capability on a smaller platform. There are even US startups to make fully US-produced embedded AI platforms like Jetson but cheaper and military oriented.
@@Thaidory this isn't anything like you said though. They dont auntonomously find and kill humans. They have target recognition and lock on. The flight coordinates are chosen by humans and so is the decision to strike. Target recognition and lock is a thing since the 80s, originaly in fighter jets.
@@Thaidoryyou said autonomous drones deployed from both sides. Everyone know anout lancet drone, yes it takes off inside of a Russian territory and looking for target on ukr territory. Is that you mean when you said "both sides"?
@@愛を込めてロシアから No. I mean both sides use smaller and cheaper FPV drones that have the same capacity as Lancet to fly into designated area of operations and engage targets on their own using machine learning.
The latter type of drones evolve fast.
Given that France has moved to the two seat Rafale since it has proven more effecive, I submoit that it might be useful two have the future a/c be two seaters.
Rafale is the only omnirole plane that can do multiple missions in a single sortie. Other multirole aircrafts can do multiple tasks but do just 1 per sortie. Pilots are trained for a specific task and not multiple so they still send a pilot trained for that sort of mission. So rafale being an omnirole, they have 2 seats so that there is a co-pilot that can focus on another task as not to waste the omnirole capability.
Also would be hard for f35 i think it would mean less internal fuel
Imagine.
BLUE ANGELS WITH NO PILOTS!
The AI will be used to make a discovery to bring people along and still perform, either some kind of remote link with your brain, or some kind of way to nullify G-forces, or some new propulsion that isn’t affected by G-forces.
science fiction at the moment and then i would not trust america with that technology
30 Years...its already here...
Re: spectrum
I would suggest looking closer at the economics of say, Geran and late model Lancet drones vs. any UCAV. If the cost ratio is literally 1 to 100 or greater, it is not clear to me that there is any real point to the UCAV from ground attack purposes.
Even from a defense perspective - again the cost ratio becomes critical. A $10 million, manned or unmanned, aircraft against 100-$100K attacking drones - the defensive aircraft is not going to do a damn thing to stop the attack wave.
Increase the $10M to $100 million - then the opposing attack force can add in some long range AD or "missile truck" air to air support for "cover", at which point the game is over.
You can probably see from the above that Russian philosophy of focus on ECW and AD is far more extensible with cheap, plentiful drones than Western high cost, low number platforms regardless of the presence or absence of pilots.
You uploaded this at the same time as the advanced Iranian drone Gaza was unveiled. Did you you know something ahead of time?
Not at all
Yeah I have to call Hoey on that.
Oh I know we are on the way out 9 G hurts my body and as sensors surpass the Mark 1 eyeball I figure you are probably right.
So on the first point I will cede, however I will not shed a tear for pilot types EGO overloaded this is no problem (BOOM)
We will still have E G O ' S
Great video. I believe we are missing the fact that manned aircraft like the b21 can be Omni role aircraft. The b21 if you just tripled the numbers would come down in price to half the current price and armed with say twelve aim260 and 2 lrsam could dominate a huge area of operations . It carries radar and if you added thermal sensors two of them could overwhelm anything out there. Flying say sixty miles apart you would passively lock the target never using active radar. Plus you still can attack a ground target
so... i think turing test will soon be broken but AI will still not think...
Machines will follow there programming, so this makes them predictable. Being predictable can be something to exploit by your opponent.
I also doubt if smaller countries will be interested to buy a fully autonomous weapon without full access to it's programming.
There is no guarantee that those aircraft won't be used against there foreign buyers or be rendered useless.
This is something that would make foreign sales of these systems impossible.
black box analysis can solve that
come insegna BSG, se un aereo AI viene distrutto, qualche secondo prima potrebbe inviare tutte la sua exp operativa e la situazioni che lo hanno portato a perdere in un cervello quantistico, in qualche base o nave madre, che rielaborando la situazione troverebbe in pochi secondi/minuti, una soluzione ed aggiornare tutti gli altri velivoli in brevissimo tempo. se non addirittura in tempo reale.
una aviazione del genere che si auto modifica ed ottimizza è praticamente invincibile nel medio lungo periodo. è anche vero che la troppa ottimizzazione ti garantisce solo risultati ottimi nel tracker passati e non in quelli futuri..
detto questo c'è una falla operativa. se uno disponesse di un computer quantistico superiore potrebbe hacherare questi velivoli e quindi tutta la tua forza ritrovandotela contro.
penso quindi, che l'uomo in definitiva è insostituibile. una macchina per quanto evoluta e precisa resterà sempre una macchina. utile per fare la macchina. mentre l'uomo è la coscienza che la dirige.
non sono quindi per una IA auto cosciente. ma per una al servizio dell'uomo.
quindi una integrazione piuttosto che una sostituzione.
Top Gun 3?
I'm a big non-believer of AI!
But let us assume that the highest level of autonomous aircraft exist. But what will they fight against? Air defense on the ground? No, because either the drone is clever enough to realize it has no chance, or the ground forces are clever enough to realize they have no chance. Will those drones be heroic enough to fight against superior drones of the enemy?
All this is so much science fiction that it will take us a century to understand the complexity.
AI: If our brain would be simple enough for us to understand it, we would be too simple to understand our brain.
The weight tradeoff is compensated for by the vulnerability of AI pilots at the system level to countermeasures ranging from EMP to spoofing to what I call response spectrum.
Response spectrum is that every human pilot in any given situation will not act the same, but every AI pilot will. A single successful attack on the AI (i.e. control system) or a bug or any other serious deficiency - will be present in every single AI controlled plane. This is a point failure that becomes a systemic vulnerability due to ubiquity.
There are also very serious questions about cost and adaptability: the spectrum performance of AI comes about via massive parallel testing in data centers; significant, much less major modifications of the base "AI" behavior require a rerun of a lot of the CPU expenditure in the development process. This is obviously not optimal in a highly reactive warfare situation. This process is also extremely expensive - so much so that it is debatable whether AI pilots are actually cheaper than training real pilots unless the size of the air force is over a certain point (i.e. small air forces are better off with human pilots, from a cost perspective).
Or put another way: the scalability of AI pilots into multiple platforms (i.e. copying software over) is also its greatest weakness.
there are actually some good tricks you can use to re-train a deep nerual network much more quickly than you trained in initially. In computer vision for example, you can often take a system are retrain it to look for something completely different (assuming the same input format) just by retaining the last layer or two, which are usually also the smallest.
I'm pretty sure the ability to only partially retrain is a feature of all deep neural networks, so that helps a lot with responsiveness. You can also add a little more noise and tolerance into the training algorithm so that you can easily produce a family of really good solutions instead of a single perfect one, which would help keep things less predictable.
if you want to make the system more modifiable at the cost of a little performance, you can make it out of a few separate platform-specific and mission-specific AI systems working together that you can replace one at a time.
Pilots in cockpits arre a performance limitation.
I think you neglected the importance of ultra low cost drones like the FPVs used in the Ukraine war. They are even being used to assassinate single soldiers... So why build very expensive aircraft, ships drones etc. when you can build assassination drones? I think autonomous assassination drones are by far the most cost effective weapon.
For other types of drones I think mini cruise missiles with ~20kg HEAT warheads are also the future. One would cost on the range of $20-30K but be able to destroy light infrastructure, most vehicles while at the same time have massive range (thousands of km).
For all intents and purposes FPV drones are simply guided ammunition. M*7 was talking about entirely different class of drones.
@@phunkracy Both Russia and Ukraine are actively developing autonomous versions of them.
Have been waiting for drone swarms to get displayed in the Ukraine war. Just dozens of drones flying across and obliterating positions like a computer was doing it.
And yet it didn't happen...that is an interesting point.
Oh God they'll have to get real jobs