Russia-Ukraine: Simplicius the Dipshit Finds One Weird Trick to Revolutionize Warfare, Let Easily Spoofed AI Make Life and Death Decisions

I have been waiting for the Russia-Ukraine war to end so I can re-adjust my initial quite incorrect prediction of an almost immediate total Russian victory. Taken from that article.

The Ukrainians had years and years to build up such a system, they just didn’t. As a result their air force is destroyed, and the rest of the war is an afterthought. Russia has already completely won. I don’t push in my chips at the clout casino lightly, so consider that a fact.

In reality the Russia-Ukraine war has been a long, slow grind for both sides. Finding out exactly what is going on is truly impossible. The WMD Liars are willing to lie about many transparently idiotic things. Getting casualty numbers from them is next to impossible. What is definitely true is that my prediction of Russia simply steamrolling Ukraine in a matter of weeks has undoubtedly been proven false.

Distracting from this was the utterly delicious Reddit Dilation Brigade getting absolutely deleted on the battlefield. This was great fun to watch, but distracted from the reality that the overall war had turned into a grinding stalemate.

I’d suggest watching the above video from the time I started it on 2x speed. They start with the Donbass War, and at around the 4 minute mark proceed to the Russia-Ukraine war. While we can quibble over the details of the map, it’s clear that there were enormous Russian gains up until April, followed by a retreat back to the Donbass, followed by almost a full year of very incremental gains for both sides. Currently the frontlines look something like this, with the overall map at the bottom, and Bakhmut taking up most of the area.

So what happened? Why did we not see a Russian blitz reminiscent of the Battle of France?

I am quite convinced that there is no one singular answer, but if I had to point to one fixable thing, it would be the absurd reliance on attack helicopters. I pointed out in my Fighter plane series that you can’t use jet fighters to do CAS. They simply have no ability to find things on the ground in the first place. Attack helis can, except that they are so fragile and terrible that they are essentially useless.

Because helis must lift themselves with propulsion, instead of with lift generated from wings, they have terrible range, terrible endurance, and terrible payloads. They also have very little ability to pull hard Gs when turning, which is essential to survival. They make terrible weapons platforms, since they can’t actually hover while firing weaponry, aside from very light small arms miniguns fired out the side. Finally, they make tremendous amounts of noise, which gives their positions away, and can’t fly at high altitudes when that would be safe.

Attack helis failed in Vietnam, then failed in Iraq. They appear to be in the process of failing in Russia-Ukraine, with one and only one saving grace.

Above we see a video of a Ukrainian attack heli firing dumb rockets at Russian positions. Or at least that’s what the video purports to show. The Ukrainian Air Force has resorted to the same attack heli cope as the Russians: lofting rockets and leaving. I’ll get into that in the next piece.

It is of course an accomplishment that they even have helicopters at this point in the war. This is because helicopters can takeoff and land from almost everywhere. This means that, unlike their jet fighters, which were destroyed on the ground very early on in the war, they can still operate their helis as long as they can hide where they operate from, and keep them maintained and fueled up.

That’s the problem with planes that require pristine runways and airbases to takeoff and land from. They’re absolute jokes in a real war. Any plane that cannot takeoff and land from rough, barely prepared airfields is not a serious war vehicle. It’d be like a tank that could only operate from pristine roads.

But there is a reason why helis have traditionally been used as troop transport vehicles. Their ability to go up and down makes them great a picking up troops. You simply cannot replace that with some STOL fixed wing aircraft, you need helicopters.

Unfortunately, they are extremely fragile. The above video is mislabeled, as we clearly see a troop transport helicopter being downed, but there are plenty of videos of attack helis being downed as well.

And here are some Houthis shooting down some Apache helis that the US sold to the Saudis.

One often overlooked problem with helicopters is also that, for practical reasons, you can’t put an ejection seat on them. This isn’t so bad for troop transport, since we care about more than the pilot, but for attack helicopters it’s a serious drawback.

Both sides of Russia-Ukraine have nothing comparable to the above OV-10 Bronco, which would be getting serious work done. But rugged Close Air Support or Armed Air Recon planes are cheap and unprofitable, something neither our nor the Russian Military Industrial Complex is interested in.

One other type of aircraft seeing use on the frontlines are quadcopters. I took a giant dump on long range UAV’s before, while making sure to say that short range line of sight UAV’s – the kind you can buy at the local hardware store – do provide niche value. They can be launched from just about anywhere, albeit with terrible range and almost non-existent payloads.

Today we see an analysis which once again overhypes drones into the stratosphere, along with a whole host of other nonsense involving AI and autonomous systems. The first half of the article is a passable history of some warfare. Then we jump into this.

Simplicius:

Today in Ukraine we are witnessing what is likely to become another epochal shift—a passing of an era. The beginning of the conflict saw clumsy utilizations of first-generation drone technologies, like the basic family-friendly DJI Phantom models. This quickly expanded into the smaller and more versatile DJI Mavics, and other type of larger and more powerful octo- and hexacopters, fixed-wing-VTOL-hybrids, and now VR-goggle-aided FPV racing drones.

Drones began to be modified with special batteries, receivers/transmitters, and propellers to expand their endurance, range, and killing potential. Equipping them with Thermal (IR) cameras likewise became ubiquitous, lending drones the full-spectrum ability of near total ISR.

But even so, the clumsy grasping of the early stages of this evolution becomes apparent in light of future possibilities—but has revealed hints of the coming storm of advancements which will serve as the ‘bifurcation’ point from which the conflict will never look back.

Kill me now.

Let me explain the actual utility of these short range quadcopters. You stick a camera on them and send them up into the air, because you can get a really good look at things with just a bit of altitude. While it is not crucial to have them, and they are susceptible to being shot down or crashing due to losing signal, they can provide a nice benefit to the ground forces. The vision that a camera will give you is not good relative to an airborne pilot, and the endurance and range of the things are terrible, but they have their time and place.

For the record, an octo-copter powered by a gas-electric system set a record for endurance of 13 hours, and traveled a distance of 205 miles. This is a one off science project though, and in reality drones that have almost one hour of endurance are marketed as having excellent endurance.

Quadcopters that can fly for even thirty minutes, with realistic configurations weight, are considered to be doing very well. In the below video the guy explains how he’s doing well with a standard drone to fly it for twenty minutes.

The problem that quadcopters face is the same problem that helicopters face, combined with the problem that electric vehicles face. Helis have to keep themselves upright with power, which is very power hungry, and batteries have terrible energy/weight ratios. We covered this with the utterly useless Tesla Semi.

The engines and propellers used to keep them aloft, as well as the batteries used to power them, come from very mature industries. They aren’t going to magically get better, no matter how many South African fake engineers mumble about random gibberish. 

The terrible endurance of quadcopters isn’t getting significantly better. The terrible payload, and therefore camera quality or weaponry, isn’t getting better. The terrible range isn’t getting better. And there are serious issues with any communication to them being jammed, or being accidentally blocked due to trees or anything else getting in the way. In short, they are a niche system.

Not understanding this would be fine. It’s this next part that really gets me.

And many thought-leaders on the Russian side, too, have begun calls for a re-orientation towards future systems. Dmitry Rogozin, who’s previously served as head of Roscosmos (Russia’s ‘NASA’) and Russia’s defense industry, recently posted such a call-to-arms. In his very first paragraph, he echoes our own recent sentiments by decrying the use of combat aviation in the modern ISR-rich battlefield, which has, in essence, obsoleted the glorious ‘strike hammers’ of the 80’s.

“The battle with NATO and its puppet and heavily armed Ukraine showed that modern war is a war of robotic means that ensure the effectiveness of artillery and assault infantry. This is a war of avatars, combat robots, when the outcome of the battle is decided not by a two-meter giant sollate with a formidable machine gun at the ready, but by a “smart bespectacled man” who creates a reconnaissance and attack UAV with a protected radio link, a complex for overcoming enemy electronic warfare jamming and localizing his UAV according to video data, as well as another “smart bespectacled man” who can deftly dispose of this means of modern war, establish the location of the enemy and within a couple of minutes give our art the exact coordinates of the enemy.”

Funny, we had that exact thing before. It was called Forward Air Controller, and it was done by planes that look like this.

I guess this is magically better when you don’t have a guy in the UAV for some reason, instead having to rely on a complicated and fragile network of relayed radio messages for your artillery to be properly guided. That way you can have unreliable artillery, in exchange for never having a guy die while doing Forward Air Control. Because that’s the point of the military, to not have pilots die while doing FAC.

He goes on to outright call for the entirety of Russia’s aviation and fleet capabilities to be unmanned and autonomous.

Military Industrial Complex fake engineering is just the boomercon version of what the traffic antifas do.

Similarly, it is time to finally understand that both aviation and the fleet should be predominantly unmanned and have increased stealth from the enemy and autonomy of application. Air and sea drones will inevitably displace combat aviation and the traditional fleet. And all other participants in the armed conflict should proceed from the fact that all of the action will be fixed by means of optical-electronic, radio-technical and other types of enemy reconnaissance – space and unmanned / air, and the most important means of their survival on the battlefield is the maintenance of “indestructible” communication between units and the efficiency of decision-making.

It should be noted that both the new Su-57 and T-14 Armata tanks were designed with un-manned capability in mind. Future versions were meant to be pilot-able remotely, and Russia’s newest S-70 Okhotnik drone is likewise meant to eventually have an autonomous ‘wingman’ mode for assisting Su-57’s missions.

Unmanned tanks?

“As soon as deliveries of Abrams and Leopard tanks to the Ukrainian troops begin, the Marker will receive an appropriate electronic image and will be able to automatically detect and hit American and German tanks with ATGMs.”

Machine learning for tanks. We’ll get to why this is an astonishingly terrible idea later. For now, just know we can’t even get autonomous driving cars working.

He goes on to uncritically bloviate about Artificial Intelligence. 

But beyond that, A.I. is the most ascendant technology which will soon transform the face of all conflict, including this one, should it last long enough. A.I. systems which can ‘smartly’ negotiate terrain and find, identify, and even engage targets on their own are already in nascent prototyping phases all over the world, and will more than likely soon see their entry into the current conflict.

AI drones on the battlefield.

Don’t worry, he shows this LARP machine from Yidsrael as proof of the concept.

This Israeli Elbit Systems drone for instance, can autonomously negotiate interior environments and use facial recognition to detect and engage enemy combatants on its own.

This is almost too stupid to debunk. Enemy soldiers are going to be camoflauged. Their faces will be painted. They will be hiding. You have no way of finding them in the first place with this LARPster, and no way of actually identifying them even if you do. Good luck with friendly fire accidents using this garbage. 

For those that may balk, this technology is not revolutionary. Social media apps and iPhones have already long used facial recognition algorithms to detect faces in photos. Consumer cameras, also, for several years now, had facial detection autofocus features, allowing them to automatically ‘focus’ the lens on a person even if they are moving back and forward. Hell, even cheap new consumer security systems like this one from ADT, now come with facial recognition which saves faces of your friends and family and notifies you when they’re approaching your home, and conversely can notify you if it’s a ‘stranger’.

Great, now all we have to do is have the enemy soldiers line up and show our drones their faces. Remind me why we aren’t just using artillery again?

Google has been pioneering this work under their contentious ‘Project Maven’ banner, which elicited a near-mutiny of the duly worried employees, for the grisly military implications such technologies entailed.

Just last week, this new exposé showcased how Eric Schmidt, ex-CEO and Executive Chairman of Google and Alphabet Inc., has been working hand-in-glove with the U.S. gov’t to accelerate and streamline their AI technologies acquisitions and development. In fact, he left Google to make this his new career and full-time pursuit. He is apparently beyond driven as regards defeating China in the AI war, and goes to great lengths to underline the point that AI is the technology that will determine the future’s winner.

He quotes from Mr. Schmidt here.

A New Weapon

The Pentagon’s tech problem is most pressing, Schmidt says, when it comes to AI. “Every once in a while, a new weapon, a new technology comes along that changes things,” he says. “Einstein wrote a letter to Roosevelt in the 1930s saying that there is this new technology—nuclear weapons—that could change war, which it clearly did. I would argue that [AI-powered] autonomy and decentralized, distributed systems are that powerful.”

I could not agree more that distributed systems are powerful. Please read the second part of my fighter plane series where I show the serious limitations of fuel for determining the maximum size of the military. I suggested that we stop building massive refineries which will just be destroyed by long range ICBM’s on the first day of WW3.

But apparently the MEGA GENIUSES understand that having easily destroyed infrastructure is totally fine. Instead you just have a bunch of shitty little drones with magic LARP technology that something something distributed systems something something lethality something something force multiplier. 

There’s a multitude of ways in which AI will very soon revolutionize warfare. Many of which are already in their inaugural stages, others have progressed much farther along. Those who’ve read my ISR article know the growing importance and dominance of that field; but now imagine in place of human analysts laboriously poring over endless gigabytes of satellite data to locate Russian troop positions, instead a tireless, and far faster and more efficient, AI algorithm scans and processes thousands of hectares worth of terrain data in seconds, picking out every single targetable object of interest, sorting and collating them into appropriate baskets, and even—eventually—autonomously routing the data to the exact, appropriate sector fire unit which the AI judges to be most capable, ready, equipped, etc., to handle the task.

Once again I’ll get to the AI nonsense later. But for now you will notice that he’s assuming the bottleneck is the number of human analysts who can look at the satellite images, and not the satellites themselves. This is absurd. Satellite time is extremely precious. We are not running out of analysts to look for likely enemy sites. So this is optimizing a part of the problem that doesn’t exist.

And once again, satellites are only good for surveillance. They can’t find things, because they look at such a tiny amount of the Earth at any one moment in time. 

One might think radar to be fairly simple: point the beam into the empty sky, whatever pure signal is returned is the target.

But here’s the catch: in today’s ISR dominant environment, almost EVERY asset of note flies low, “under the radar” for the sake of evading it—from combat aircraft, to drones, to even cruise missiles which are now exclusively programmed to ‘hug the ground’ via terrain-mapping.

This is undoubtedly true, although it only contradicts other bullshit of the military industrial complex, and does not argue in favour of AI drones, or in this case AI radar. 

Thus, the importance of ‘Look Down’ radars rapidly becomes supreme. These are radars fitted on things like AWACs which can scan not just the sky, but the ground, and differentiate a moving object against the vast clutter of other ground objects like moving cars, people, swaying trees, etc. For a computational system to distinguish a helicopter moving at roughly car-speed at 50ft above ground, from other cars traveling directly beneath it, one needs massive amounts of parallel computational power feeding into advanced AI algorithms which can very ‘smartly’ distinguish this data. This is how such radar systems can even detect ground objects like tanks and mobile formations on the move.

Contrary to the bullshit of the Military Industrial Complex, the way to hide from radar is not to build fake stealth fighters, which can be picked up on VHF radar easily. The real way to hide is to fly low to the ground. But the higher the lift/speed ratio of the plane the easier they can hug the ground. That’s yet another argument for having jet fighters with good takeoff/landing speeds even beyond rough field takeoff and landing.

Here’s what’s left of the Ukrainian fixed-wing aircraft doing the same.

The way to survive is not to fly super high and hide from “muh MANPADS.” The way to hide is to fly super low and hide from everything, including radar. This is odd, since that’s exactly the opposite of what I was told from shills trying to sell this garbage everywhere.

Like always, planes should be designed to excel in the ways that they are actually used. This is a strong argument in favour of straight wing prop driven airplanes using a real form of stealth, low sound emission coupled with low Infrared emission.

Above we see the Lockheed Quiet Star. It was the quietest manned plane ever built, supposedly as loud as a whisper when flying at 500 feet overhead. It’s not hard to believe considering the quieting that people do nowadays with quadcopters.

Above we have a video of the quadcopter in normal configuration, started at the appropriate time. Below we see it after it has been outfitted with heavier, but quieter propellers.

Muffle the engine of a prop driven OV-10 Bronco and add two quiet propellers (which are a bit heavier and less efficient) and you can enjoy one tenth the sound production. Couple that with a slower and smaller plane and you’ve got a real stealth plane in the real world. Not one that is designed to do raids on the taxpayer, but on the enemy you’re fighting.

Sadly, there’s still just no way to make obscene profits off the thing. I guess we’ll just have to stick with garbage that doesn’t work.

Make no mistake, once the Russia-Ukraine war is over, the MIC will immediately forget that everyone started flying low to the ground to avoid radar and SAMs, and continue talking about how low altitude planes spontaneously combust due to MANPADS in the ninth generation Integrated Air Defense Systems of the fortieth generation future battlezone due to sensor fusion.

But the final frontier in AI capability won’t be simple algorithms for ‘smart’ detection of objects, but full on autonomy scaled up into network-integrated assets which can communicate with each other, and solve tasks together. In layman’s terms, this describes the coming age of ‘drone swarms’—which is by far the single most critical area of development. And here, China appears to lead the field.

Recently released footage demonstrates a swarm of autonomous Chinese drones navigating a bamboo forest without the use of GPS.

I can’t take this bullshit any longer. Let’s skip ahead.

We’ve established that those who have the biggest computational power will have the most advanced AI systems, with the most possible pure instruction/operations-churning abilities. And that means, those with the best chips, i.e. semiconductor industries and capabilities, will be the kings of the coming AI wars.

This also makes no sense. Russia has enough chip manufacturing to produce the tiny amount of extremely small CPUs to run the very minimally demanding software that real technology, like GPS guidance or fly by wire, demands. Even machine learning requires large amounts of compute power to create, not to operate after the fact.

One day we’ll look back at these dim birthing moments of consumer-grade drones crudely hoisting bomblets into trenches in the same light we took to that seemingly lawless, antediluvian world of aerial pistol shoot-outs, long before the Red Baron’s Fokker scourged the Allied skies. And with the galvanized national spirit, the unprecedented solidarity of the Russian narod, and fervor of ingenuity seen daily in their fighters, it’s clear that Russia will be the one who takes the reins—and leads the world by the hand through the inchoate darkness of this new era.

With that it is mercifully over.

Almost immediately we see the following comment chain. Just skim the bold.

Commenter karlof1,

Back in the 1960s, we built rather large scale model powered airplanes, so the tech to run drones has existed for decades. What deterred it was man–he had to be in control. Now we have the world of the Terminator on the horizon since nobody ever thought to enact Asimov’s Three Laws of Robotics made in the 1940s. Jets can do far more outrageous acrobatics unmanned–hypersonic fighter planes. Droid Armies. And yes, the nation having the most capable system of promoting its human capital so all brains have the opportunity to contribute will be in the vanguard. Russians, Chinese, Iranians, and others have known that formula and are very motivated to see it succeed and thus support it.

You already know my view on drone usage. The next development IMO must be an electronic cloaking device to makes the object invisible to the sensor suite, otherwise human infantry will eventually vanish to be replaced by droids. Then we have the resource issue becoming what nation has the resources to produce combat droids in vast numbers. And I doubt those droids will look anything like those from Star Wars or anywhere near anthropogenic. So, yes, your premise is sound and being proven as I type. I have no problem imagining all sorts of different engineer droids made to deal with fortifications that would replace human infantry.

That was a random guy. Having schizos in your comment section is not to be held against anyone. However, Simplicius agress with him. 

Simplicius response,

You’re right,

and to some extent they already have it — for instance Russia’s Kh-101 cruise missile which has been the bread and butter of all the recent cruise missile strikes on Ukrainian infrastructure, is said to have a EW jamming suite which has already seemingly been witnessed screwing with AFU air defenses. There were a few videos showing Kh-101’s passing by towards Kiev, and the Ukrainian AD missiles flying right past it or strangely whizzing off in a random direction nearby as if they were suddenly ‘spooked’ by something electronic.

The smaller drones are too small to have such powerful suites simply because typically EW jamming is a byproduct of electrical power, i.e. the more power generated the more powerful the signal. And these huge cruise missiles with their turbine engines can generate enough power it seems to produce a strong counter signal, but small drones with their flimsy batteries can’t do this. So it would likely take much larger drones to be able to produce strong enough jamming signals.

Reader Thomas Taylor manages to talk a tiny amount of sense. 

Hate to break it to you, but any missile generating a jamming signal is broadcasting their position. That’s not what you want in a cruise missile.

Your anecdotal BS is misleading your readers.

I’ve had to deal with this garbage so much with the fighter plane series. There is no way for anything to send out signals without being detected. The whole point of a cruise missile hugging the ground is so that they don’t get detected in the first place. The last thing they want to do is broadcast their position. And all of this “smart missiles” stuff is nonsense anyway, since the real tactic is just to fire dozens or hundreds of missiles at the target and saturate the defenses.

But it takes until user c1ue to pop the AI balloon. I’ve transcribed this below.

Consider me an AI skeptic.

It is extremely easy to fool AIs with even rudimentary camouflage much less full Maskirovka.

Yes, machine learning is retarded.

Imagine relying on this thing in an environment where the enemy is already trying to fuck with you constantly.

There is also all manner of ways to poison data feeds.

Combat recognition using AI is an invitation to disaster.

As for chips: I have direct personal career experience in this area. The chips used in combat gear are old ones – decade to 3 decade. They aren’t expensive, they aren’t rare. It is ludicrous to think that iPhone dominance means anything whatsoever in terms of semiconductor impact on combat capability. Let’s not forget that combat gear has to be at least somewhat durable: temperature highs and lows, vibration, electromagnetic attacks or just noise, inconsistent and dirty power supplies, radiation hardening, etc etc. That which is used to build civilian systems has no place whatsoever in combat gear.

Yes, there is a role for semiconductors in non-combat gear such as aforementioned radar AI – but again anyone and everyone attempting AI use in combat is inviting massive risk.

Think of the difference between a car on cruise control vs. full self driving: the former is well bounded, easily understood controllable assistance to an experienced operator while the latter is a multi-billion dollar boondoggle which has plateaued in its development – and this is in an environment without an active opponent. I suggest looking at the project which drew out what so-called AIs actually saw in image recognition – it was eye opening in a bad way.

All of this is good common sense. User Flabbergaster joins in,

You’re not the only AI sceptic here. The first thing that came to my mind was how when personal computers were introduced into the office, all the ‘experts’ declared the advent of the paperless office. The reality was of course that printers came along with the PC’s and the demand for paper skyrocketed. This alleged AI revolution will turn out similar to that, in my opinion.

AI/ML doesn’t spot tanks, or people, or artillery pieces, etc. It spots objects, shapes and configurations it has been programmed to spot. That’s it. That’s what it does. It doesn’t know what a tank is. It doesn’t know what a person is. It doesn’t know what an artillery piece is. And it’s pretty easy to fool.

Indeed. The above neural net thinks it’s looking at a toaster. It’s undeniably ready to be shoved into a combat zone with lives on the line and civilization hanging in the balance.

Back in ’99, when NATO launched it’s war against Yugoslavia, it turned out to be extremely easy (and cheap!) to fool NATO’s high-tech sensors. These too were supposed to spot tanks and artillery and such. But the most realistic estimates are (by NATO officers!) that during those six weeks or so of bombing, only about 14 Armoured Fighting Vehicles of all types, and between two and three dozen artillery pieces (from 81mm mortars upwards) were destroyed in Kosovo. It was a pitiful result.

One of the methods used to detect targets by NATO was shape recognition. So the Yugoslavs painted black tank outlines on the road surface. And paintjob after paintjob was then destroyed by very expensive NATO smart bombs and missiles. When NATO started worrying they might be targetting hulks that were already knocked out, they started to use heat sensors, figuring that in active vehicles there would be temperature gradients visible (from a recently turned off engine for example). How did the Yugoslavs beat that? They simply put a big tub of water on the part of the painted on tank were they engine would be. The water and the road surface held on to the heat of the sun differently, and once again expensive armaments took out painted tanks. And tubs of water this time. So there was some improvement on NATO’s part.

And it’s not going to be any better now either, with ‘AI’ on the contrary. More computational power coupled with easily fooled sensors means even more ‘painted tanks’ destroyed. And to be clear, it’s not the tech that needs to be fooled or tricked, but the very human assumptions underpinning the tech and its programming. And those are not nearly as sophisticated. On the contrary, in that regard there seems to be a tendency to regress rather than improve, as people put more and more false faith in the technology itself, and progressively undervalue the human factor.

Recently in the US they tested some AI driven tech on its ability to spot and engage approaching enemy soldiers. They challenged a bunch of marines to try and come up to the ‘AI guard’ without being ‘engaged’. It failed miserably. Worse, it failed to engage even a single marine who approached. The marines used a wide range of tricks to fool the AI. One skipped the whole way. Apparently ‘skipping’ wasn’t included in the programming of possible modes of human movement. Another one just hid underneath a carton box and crawled his way forward, and said to have been giggling incessantly. And there were plenty of other ingenuous methods used. I think one dressed as a tree.

This actually works against Neural Net AIs…

One of the things that will accompany this use of AI will be a massive false sense of security. And at some point it’s going to come back and bite someone in the ass big time.

Yes, this is real. They asked a bunch of Marines to try to advance on a position guarded by a system of cameras interpreted through a neural net. They all made it. Some of them did cartwheels over. Some of them skipped. Two of them just hid under a cardboard box and pushed it forward, in the real life Metal Gear Solid. 

Another guy just pasted a bunch of tree branches onto himself and the AI thought he was a tree. Or rather, it didn’t think anything. It’s just a program that recognizes patterns it’s trained to recognize, so it didn’t think anything of the tree moving towards it, because it doesn’t know what anything is.

And no, the more things you train the AI to look for, the worse it gets at understanding any one thing in particular. Neural nets will always have false positives and false negatives. Exposing them to an environment with a hostile enemy is the height of idiocy and lack of understanding.

I’m just so sick of living in a world where people feel entitled to spew utter garbage totally divorced from reality. Then they simply move on to the next thing when it’s proven false. Gradually, I began to hate them.

In the next piece we go over Russia’s very poor air interdiction capabilities shown throughout this war.