Finally someone (Ziga) addressing the elephant in the room: that the current "AI" is in fact just a virtual "intelligence". We are far from real machine self-learning and self-awareness.
AI is just that, Artificial Intelligence. Being self-aware is not part of the criteria for what makes an AI. That's only a thing in movies. Machine learning is already here, and extremely common. In fact, if you were recommended this video, it was likely the result of machine learning.
Not really that much of a elephant in the room. There is going to be a fine line very soon between what is virtual "intelligence" and what is just intelligence. We may have created it, but that's a moot point if it keeps exponentially improving. There is a reason the tech world is obsessed about "the singularity" because that's just how exponential growth curves work. Maybe human level or greater general intelligent AI is 100 years away, or 200 years, or it may be 5 years away. The honest truth is, even at top level research and universities, we don't know, it's allot of guess work and assumptions being made, there are many very qualified and very smart people that think we are closer to reaching that goal in 5 years than 100. People would have laughed in your face not to long ago about everyone having a super computer in their own home, now we have super computers in our pockets, hundreds of times more powerful than the entire computer system used in the Apollo moon missions. Even old science fiction didn't go that far in their predictions. A point about exponential growth that even very smart people have a hard time grasping. If someone asked you what would you rather have, $500,000 or 1 penny today that doubles each day for 1 month, the vast majority of people would take the half million, yet starting from just one cent and investing it two-fold the next day, by the end of thirty days, it would grow exponentially - to a massive $5,368,709. Self improving AI has a potential for exponential growth, and allot can happen in a very very short period time. We also often forget that for thousands of years humanity went from 90% farmers and riding horses, to industrialized society, supersonic flight and traveling to outer space in the span of a single lifetime, and technological progression is still rapidly growing.
@@LN997-i8x I could argue that the criteria is incredibly low. But that's a philosophical view, I know. One could say that "if sentences" and "if programming" does not constitute real intelligence but then again, even humans are "programmed" through life by the same principles and you can fully understand one human being through "if matrix". And it's the same with abstract thinking. Abstract thinking is how we define in animals which is intelligent and which is not. Is the current AI really abstract thinking, if it has pre-programmed answers? Then again, is a human?
@@-Zevin- "elephant in the room" from a marketing perspective. On one end you have what the scientific community is doing and on the other you have "if algorithms" being marketed and sold as AI. And now everything is AI, from your coffee maker, Alexa and whatnot.
@@perunikavarnost6472 Yeah I can't disagree there. Just noticed your name too Перýн, pretty cool, don't come across the Slavic pantheon very often. I had a axe of perun myself cast in bronze from a old archeological find, wore it on a necklace for ages.
I'm an American doing a semester abroad in the Czech Republic and it's been a mindblowing experience. I've never been to Europe before and have been amazed by how different everything is. I plan to try to make a trip to Slovenia at some point.
@AlexLee-dc2vb if you are in CZ certainly stop by the store on Opletalova Street in Prague. Guys there will give you some recommendations what cool ranges to visit. :)
Seeing this podcast is giving me flashbacks already. It was a pleasure to compete with you guys at Lynx Brutality 2023. Can wait to see you guys again.
I think non-intelligent AI shouldn't be underestimated for firearms development: "Take this design and reproduce it with less parts to manufacture", "take this blueprint and lower the pressure on the bolt face", "find a way to combine these two functions in single part with improved geometry " Those are the type of prompts that would require very little intelligence, but would take months for a human engineer to crack
Really good thinking, even just these simple commands could do a lot. But them imagine a really advanced system that can make the optimal caliber/ammo/rifle combination
This sounds less like AI and more like a very specific program similar to digital wind tunnel testing or scientific fluid dynamics programs. It's all dependent on the parameters that you put into it when you write the code, and that has to be VERY exact, to the point where it's more work to write the parameters than to do the job yourself. The program would need to know things like "yes, aluminum is light weight but also a bad choice for a barrel". It also has to know what you consider an "improvement". And how to weight these different demands. Is fewer parts more desirable than longevity? Than user safety? And then it has to take human ergonomics into account. No, that's where the user's face goes, you can't have the bolt moving into that area, silly AI! If you really want to combine two functions into one part with improved geometry, just hire a Soviet engineer who works only with pen and paper (Sudaev's recoil spring guide rod which is also the extractor). Or if money's not an option, Christian von Koenigsegg (engine and liquid tanks are load bearing parts of the chassis). Both will do this better than any computer, and probably faster too.
If nothing else it would speed up the process if you tell it things like, "Smooth out the pixelated surfaces on this enlarged image." "Fill in the negative space."
I feel like Ian's point about that we have already tested so many systems and we are left with the ones that worked the best is a bit premature on what has been tested. It kind of draws the conclusion that what has been tested is essentailly every possibility and we are left with a handful of sytems that work well. In reality I have the feeling it's more like the designs we have today is a fish bowl, and what we have tested is a swimming pool. But there are entire oceans outside of that that we haven't bothered with because the inter-disciplinary distance is too vast or complicated to justify the time/cost. And imagination isn't limitless, there are limits and good useful imagination is only bestowed on the few, unfortunately. What we're seeing from AI is a way to sort of bruteforce our way to a breakthrough by testing every possibility (instead of waiting for someone to connect the dots once in a blue moon), and knowing what it may look like. A proper AI would not need to draw borders between different disciplines, and will walk between them easily and utilize every tool and discovery humanity has amassed, and even come up with its own ones as Ziga mentioned, and focus it into a single goal. That was a bit long winded but I hope my point comes across, AI is not limited like humanity is and the ideas and frameworks we build up for ouselves.
Bingo! The designs that we have now are the ones that are logical to use because we know how they work and we were successful to mass produce them relatively cheaply. But this is far from all the possible systems - some that could be vastly better and more efficient, but we would need to look outside of the box. Similar to what Elon Musk did with space travel - reusable rockets were "too expensive, too complicated and too expensive to make" - until the madman did it and revolutionized the space travel. Now it seems logical to us that reusing rockets is the way to go because it 100x to 1000x cheaper than building new rockets everytime. But until someone did it successfuly, everyone thought it was an impossible idea So yes, that's one of the points i was trying to make, with tghe correct parameters and a defined goal, AI could probably come up with very interesting and efficient ideas.
electric weaponry is the future. the theoretical limits are SO far out, that laser and rail weapons (or coil) are possible even within what we can reach today! (far more interesting, once we break the fusion reactor technology barrier). smart weapons will be the next step. tiny gyros correcting for you shaking etc. smart bullets that can correct themselves etc. if you like, you can take a look at the alita manga for some hints of what is conceivable. one example are the "hornet bombs". tiny drones the size of a giant hornet, able to seek out your vulnerable spots (like the neck) where the armor is the thinnest with enough explosive to just penetrate. no wasted detonation energy. same for larger scale weapons. you have practically three limits. - physical barrier, where penetrators are always stronger than shields at some point (due to weight and structural limits) - energetic barrier, where lasers are probably the easiest thing to achieve, but basically every particle or wave than can be moved, can be used to cause a change in the target system - metaphysical barrier, this is only hypothetical, but in theory you can use quantum linkage or quantum shift to break apart the links between atoms in your target or cause fluctuations in the reactor cores etc. apart from that, the conventional ammunition will, at some point, reach its efficacy limit, due to "APS" countering. shaped charges can only dig so deep, pressure spikes can only penetrate so deep. etc. to break the limits of this ammunition, you will go towards the two barriers named above: by either going more physical (coil/rail guns, mass accelartors, orbital penetrators etc.) or more towards energy exchange (i.e plasma generators, tiny nukes or things like that).
addendum. flechettes are practically physical penetrators, just with a "flagrating" delivery method. once you can remove the energy waste and just accelerate the dart in a different way, they will come back. though technically, "bomb" shaped projectiles will outperform flechettes and "conventional" shaped bullets (small, to medium) at some point. you see that in how most super long range bullets are shaped like an offset oval, where the back end is slightly tapered and the front mostly have a flat or rounded nose. (instead of this: ^ )
@@AkiSan0its like slap rounds, or 120mm discarding sabots but in a handgun or rifle formfactor. I think 3d printing will be the only way to make that type of bullet cheap enough for general use. Some kind of metal dart like object with a hard plastic sabot.
This would be really expensive but with every "bruteforced" iteration/design that seemed among best the testing firm or company will fabricate at least 10 prototypes per design(widening the chance of failure so that we don't make a huge mistake if we only have one prototype per design but is rejected on a production fluke) and put through numerous proving tests to work out lubrication and gas pressure issues(or possibly spring strength if it used that old piston idea as opposed to Stoner's AR gas impingement) and then after averaging out actual field test data select the empirically best one(a decision based on both shooting statistics but that is only half the other is realistic physical characteristics judgement.) A very technocratic Darwinian funnel where through the top we drop in potentially infinite but realistically still alot but a good real number of prototype designs and out of the bottom emerge only fittest from the beginning survive/adapt to become the fittest. And the best part is if this is done in the private sector it will be very efficient in the name of profit unlike government jobs which all service veterans like to chip in and say feels more like bare minimum at best or incompetent bureaucratic quota checklist at worst. I don't see the state taking up the testing role anyway very rarely do innovations come from the state rather mostly adopted from the private sector, but if they did all this would be extremely classified and more taxpayer expense.
One thing that no one brings up when talking about development of new firearm technology is powder. The current smokeless powder was a huge technological leap adding more range and cleaner burns so that semi automatic guns could be designed. If we could invent some new propellent that is more powerful and efficient than current powder it would change the design of firearms completely.
I work pretty deep in AI/machine learning (happy to prove this) and as a gun guy I actually have some interesting ideas about this. Would be super interested to hear Ian's takes on this especially.
Žiga and I must have similar reading interests. I have read that 45,000-50,000 rounds of US ammunition were expended to achieve one enemy killed in WWII and Vietnam. GWOT numbers are supposedly around 250,000. Such high numbers are exactly why many countries hesitated to abandon bolt actions even in the 1940s. One fear was soldiers would shoot more ammo than a country could produce/supply if they were issued semi or full auto small arms.
There was even hesitance to the first single shot trapdoor rifles for the same reason. The British 1888 Lee-Metford had a detachable box magazine, this was highly controversial, and soldiers were only equipped with a single spare, and the primary mag was attached to the rifle by a bit of cord so you didn't lose it. They were worried soldiers would waste ammo, so really tried to drill into the men it was primarily for emergency use only, when you really need to fire and reload quick in a pinch, but not as standard operating procedure.
At 1:40 you're all talking about range and limitations of cartridges. With a couple of pals we did a very unscientific trial of 7.62x51 (147 German ball surp), 7.62x54R (STV/Norinco surp )and 7.62x39(STV/Norinco surp) at a 1200m range in Sicily. All the rifles were with iron sights, no scopes. Basically, we could make accurate shots on 80x60cm steel plates out to 700m with the x51 and x54R. After that the rounds went trans-sonic and accuracy fell dramatically. The x39 from a standard 16" Cugir AK74 wouldn't make hits over 500 with any accuracy. The other rifles used were an AIA No4 with 24" barrel and a Finnish Mosin M91/30 which is basically new (1944 Tikkasalo barrel), came from Finnish reserve stocks. Having carried the FN FAL and then the AUG in service I'd rather more ammo in .556 than a bigger round - more bullets, more hits....
You can extend the 7.62x51 up to 1000 meters with the correct ammo. Yeah, most AKs have a limit at 500 meters (again, depending on the round and altitude) as the bullet comes in transsonic ranfe and starts to irregularly deviate. I remember testing my AK at 480 meters with a new barrel, nice scope, good ammo - everything was perfect to make the hit. At 450m i was hitting the target but at 480 meters i was aiming at the center of the target but each shot hit differently around the target - to high, to low, 1 meter left, 2 meters right 😂 it was funny but also very educational to see. There are special heavy bullet designs that are made to be stabilized when coming from supersonic to subsonic flight but those are quite expensive and as far as i know, only work in .308 and not 7.62x39...
@@PolenarTactical I run a Savage BR12 in .308 with a 29" barrel. Scoped with a 12-50x56 S&B PMII. I can make accurate shots out to 1200m with handloads using 180gn Sierra heads, Lapua cases and RS50 powder...as long as the shooting gods are with me... 1100m is no problem. We had the same fun as you with the x51 and x54R at 800m, up, down, left, right...
Great discussion on xm7. I personally believe that the Army plans on the combination of armor penetration and the new optic changing infantry tactics from “suppress and move” to “see and kill”. I’m interested to see if they are able to implement this.
Having watched the whole thing, you guys really made for a fun watch. Five guys comfortable with each other and knowledgeable about firearms. Just like a night of friends shooting the shit.
I think what Ziga is missing about 6.5 Creedmoor is that it burns barrels out way too fast. That’s fine for competition and snipers, but not infantry rifles or machine guns. And I say this as a big fan of 6.5 Creedmoor.
Oh, im completely aware of this - faster projectile, higher pressures = reduced barrel life But look at the new M7 rifle, it's velocities and max pressure of the round - apparently that barrel can handle it without a problem. So if someone wanted to make a good 6.5 CM barrel that would last, they could. We have the technology but it probably doesnt make financial sense to produce such an expensive barrel for sport use because it's probably cheaper to replace the normal barrels more often
Does it, though? 556 is more "overbore" of a cartridge and 6.5 Swede and Japanese were commonly used in machine guns without major issues (apparently). The Army is also content with going with a much more potent 6.8x51.
About the issue of armor penetration, there was a level 5 plate tested that was far superior to XSAPI and the only round that penetrated it was 7.92 Mauser SMKH Tungsten rounds (Buffman test on Adept Collosus). Where 300 win mag swiss PAP and 7.62 NATO SLAP failed and custom 300 win mag. The difference is the WW2 tungsten rounds had a much larger core so mass makes a difference for armor penetration as well as velocity. The Germans duting WW2 used a powder (used in SMKH and v-rounds) that could bring greater power without the pressure being too high. The US and Soviet Union were never able to replicate it. The SMKH shoots a 194 grain bullet at 2900 fps which is wicked hot. 6.8 catridges would have too small of a core to do as much damage to level 5 as 8mm. Lethality over range is why 6.5 cartridges were considered insufficient in the World Wars. Honestly a modernized 8mm (even 8mm ball is supersonic past 1000m) or something similar would be way better than a 6.8 for destroying armor at longer ranges. A heavier cartridge would be better for machine guns and DMRs and then most infantry should stick with a lighter cartridge for closer range. Most soldiers won't even shoot at an enemy past 200m. An MG will be far more effective at range than even 10 riflemen.
The pint i think you are missing is that the large military application of the new cartridge doesn't aid the function of a platoon on fire and movement because the less quantity of ammo. There are situations where you have to spend ammo even if you have a precise optic for suppression and "stress". And the other and that at 400-600m, even with the optic, is not easy to spot people that try to hide and shoot at you.
Ian @Forgottenweapons @1:38:00 the other pro for the new rifle is barrier blind or terminal performance after going through cover. As well as eroding intermediate cover independent of belt fed support weapons. Like Ian has said the balance between issue rifle and GPMG/LMG has been contentious since intermediate caliber rifles Became standard issue.
jari is so freakin' awesome and he seems so calm & zivilised between you animals 😂 but i'm pretty sure he is the toughest dude on the table by a long shot
1:08:08 the sabots for the flechettes go in random directions potentially injuring unintentional targets. muzzle blast was referred to as "massive", probably causing issues. flechettes themselves were sensitive to any sort of barrier or wind deflection. I just read up on this lol!
I always love how chaotic podcast gets when there are so many people around the table. Sometimes it gets a lil bit too chaotic but mostly it was a blast, just like hanging out with friends. The vibe is there. :)
Play Forgotten Weapons videos - sometimes when i watch his longer videos after dinner i fall asleep in front of my screen :D His slow, calm voice is perfect for this and as an added bonus, i learn new stuff about guns
Where I think AI will be relevant in firearms is sighting systems and heads up displays. A sight with passive, image recognition based range finder, ballistic computer, and possibly later threat priority indicator for example is clearly within the reach of even our current level of capability. At the very least it can be done poorly in a lab. But that is the first step and how things like night sigths started as well.
@polenrartactical @1:48:00 I can say mortaring an X95 broke the charging handle and bent out the polymer guide for the handle. It broke the solid metal rod so maybe it was inevitable. The F2000 was not fun when I tried because the ergonomics were very different than AK/HK/AR
Another idea about AI drones: A quadcopter with a IR laser and thermal imaging. You're on a mission. At night. Dual NODs equipped. You're in Ukraine. Looking for enemies in a forest line. Your drone flies around, *looks* for targets via *Thermal* imaging - and once seen, it *points a laser at it/highlights it with IR light* *You* would give yourself away shining IR lights, pointing lasers everywhere - but if the drone shines light at enemies/points at them with laser - it's useful. You'd be equipped with a weak radio signal emitter, making you not a target for the drone (so it doesn't give you away). (kind of like how current DJIs can "follow" the user based on the signal from the controller) (signals give you away to enemy strategically, but once you turn it on within 200m of enemy positions it won't matter much)
There were complaints since Fallujah that 5.56 wasn't immediately lethal enough against certain targets. I think that perception of the troops is important. If the troops aren't confident in their equipment it has negative effects on the battlefield. Troops are less likely to charge the enemy of they don't think they can take them down reliably before they receive return fire. I know 5.56 is fine against people, but I do enjoy .308's punch more, especially against non-human targets. I'd sooner trust the 240B to stop the VBIED than my poor SAW. With my SAW I'll have to aim at the driver's face, the 240B I can aim at the vehicle's engine if I can't reliably see the driver. I wonder how does the new cartridge perform against vehicles and cover?
Take the laugo alien. Red dot with an electronic eye that can take a picture of a target. Retrofit an electronic sear. Program the sight to fire when it matches the first picture while the trigger is pulled. Would need a button to reset the target memory. Accuracy would be determined by the first shot, but the followup shots would be super fast.
Your actual prompt needs to be "how to win the war", "how to reduce suffering ASAP", "how to avoid war in the first place", etc. As soon as your prompt is _about guns_ you're basically making shooting people the goal. I think the best use for AI is out of the box solutions like the whole "intermediate calibres so it's a war of attrition against their hospitals, transportation, etc" thing (using an existing example because I kind of have to). Like Ian says, every possible _mechanism_ has kind of already been invented. Funnily enough long recoil and blow-forward are actually good solutions to questions other than "how to make the best gun". That's why the latter is all over Evan Jones's "Future of Printed Rifles" slides, where the objective function isn't "make guns better" but "make guns easier (to build)".
"every possible mechanism has kind of already been invented" this is absolutely not true and just shows how our thinking is limited inside the bounds of our successful creations and not real physical limits
@@PolenarTactical Yeah, I guess so. The reason flechette theory (speed kills, moreso than mass) works for tanks but "not" for small arms is that people noticed rifle flechettes get blown off course by wind and rain at rifle engagement distances and then went "oh flechettes don't work then". I contend that flechettes make perfect sense in the PDW role. A new fundamental machine, the rolamite, was discovered in like the 70s. That's as wild to me as a new Platonic solid being discovered. It has never been used in guns. Nic Jenzen-Jones in his ARES report on telescoped rounds pointed out that the usual psi limit in weapons is actually due to the brass brazing to the bolt past that limit. So it isn't a real, _fundamental_ limitation. Double-action triggers are never seen in long guns (except the 7.62 part of the Croatian grenade launcher), because there's no reason they're needed (in real steel), but I have pointed out elsewhere that sao sitting at full cock all of the time is a pretty bad match for printed plastic. But except for the rolamite those are all already-invented fundaments that you might use because the objective function changed. When it comes to say operating systems, it makes perfect sense to me to say that there's a fixed number of energy sources (recoil, gas, external power, a few others) and a fixed number of ways to use each one, and all have already been discovered. Same thing goes for say feed systems. The only truly new element I can see happening is electric propellant (see Integza - VERY different from either electric ignition or electric firing, which have always existed in autocannon). Then you could get closed-loop control of the burn rate, you could go subsonic based on a selector instead of a magazine change, etc.
It doesn't matter if you can hit a man at 800 metres, if you can't see him to begin with he may as well not be there as long as he doesn't see you. We also could do something like a MK12ModH in 6mm ARC and get the same hit capabilities if you slap the XM157 optic on it. It's good to have a marksman with an XM7, just like we used the M110/MK11, but even as we saw with those weapons people didn't like em. The MK11 led to the MK12 for instance. The other issue is that a massive change in equipment will mandate a change in tactic, _but_ your tactic is only viable if the enemy allows it.
I thought the new M7 was a result of learning the lessons from GWOT especially Afghanistan where the engagement range was much farther than 50-200m and was from mountain slope to mountain slope? I may be mistaken.
An AI system should be able to model and analyze more iterations of a design in a day than an engineer could do in a year. That's where I think you will see it manifest in design. You could get some funky designs though as AI common sense.
There are too many things which I wish to comment and most of I forget while listened it to the end... About AI in gin industry. Computer can't think outside the box, it only operate it's algorithm in which was included existing parts. However it can help to iterate and fine tuning of already developed idea. Just, for example simulate different weight of some parts or search ways to lightened without causing structural weaknesses.
That's the difference between a computer that you think about now and real Artificial Intelligence - even now we have really good machine learning that takes certain things, analyses them and gives us completely new and unheard of solutions (for example medical prognossis, human genome,...) Even with current machine learning and AI we only program the core of the system that then learns by itself and creates new code and new parts of itself. Look at the Chat GPT - even the people that made and trained it do not completely understand how and why it can give certain answers in a was it does
Going into the cartridge discussion on the new US rifle weapon system - the questions being asked here are incomplete, if not improper. Does a Level 4 plate change the field? No. Then defeating it is immaterial. Moving from single-shot, limited-capacity, slow-reload rifles to detachable box magazines with high capacity and quick follow-up changed the way you could conduct warfare. Rather than engaging an individual combatant, it allowed you to lay suppressing fire to deny movement while mortars, field guns, artillery, and later air support did the work. Having a Level 4 plate may defend vitals against rounds or shrapnel, but it does not allow the combatant movement in the face of incoming fire. Regardless of whether the opponent is firing 5.56 or not, the flow is still suppress with small arms and use fire support. The additional ammo capacity of a smaller, lighter round allows more intense suppression or more duration of suppression. Rather than smart rifles, you would be better off investing in smart mortar launchers. A system that can integrate GPS coordinates and optically spotted or datalinked positions for more accurate, more rapid fire. Superior fire support is the proper answer to body armor, not a small arms race.
I predict we will come closer and closer to flechettes over the years with development - this is because longer projectiles have higher ballistic coefficient, lighter projectiles can be fired at higher velocities which flattens the trajectory, reduces time to hit and can actually deliver more energy. Also lighter and longer projectiles will mean that soldiers will be able to carry more ammo
So, there is a MOSFET which operates the trigger. You have set it to double-tap a particular shaped target on the timed course. Your time runs out and you must stop, your trigger has stopped firing at those targets anyway and reverted to its default personal defence state. But before you can Show Clear it recognises the face of the R.O.
The current AI systems we have are very much AI. The "AI Effect" is a known tendency to dismiss any new AI advancement as "not real intelligence." The technology is going places, of course, but the fruits of over half a century of AI research with agent theory, action planning, and more is all around us. Everyone keeps shuffling the goalposts around.
True, but it doesnt make any economical sense to integrated such technology into simple doors. But we already have the technology as it can derive from "simple" machine learning of multiple soldiers
To your thoughts on the AI allowing you to shoot when you are on target: What i can imagine is the guns allowed for sport shooters only being capable of shooting at normed targets.
Theres those cheap shake to charge flashlights that moves a magnet thru a coil to charge tge internal battery. There are watches that motion charges the watch.
AI in weapons (in my dumb opinion), won't be in design, but in operation. Imagine a weapon that monitors target position, movement, shooter biometrics, etc and says to the shooter "don't take the shot now. Your heart rate is way up and there's unknown walking in the shot from the right"
41:00 powered rails were one of the selling points for the textron suggestion for NGSW projects. yes lots of energy is released but how do you capture that? not too many ways to do so. but you're onto something.
Instead of a buffer spring you could use a magnetic shocks, thise magnetic shocks could be involved in electricity generator. Thermo-electric generation included.
1:34:22 To Jordan’s point, in a protracted conflict like we see in Ukraine, supply shortages means artillery rounds may not be available, is being rationed, or local artillery is worried about being counter-batteried. Additionally, contested airspace could mean CAS is unavailable. In that context, an M250 with optics sounds like a pretty decent weapon.
The machine learning "AI" that we see today will affect the manufacturing side in the way Ian described it with iterating suppressor designs. Medical research is already inching towards using machine learning to help scientists trial new drugs by iterating through different chemical compositions. This could happen with bullet design as well. Machine learning AI will be able to iterate through millions of bullet shape, material construction, and twist rate designs, simulating the ballistic performance of a round, or simulating the transfer of energy as the round hits a soft or hard target. This will lead to new bullet standards. Machine learning could also iterate through countless combinations of case dimensions, gunpowder chemical compositions, or even the size and shape of a grain of powder to find the most optimal configuration for perfect powder burn. I definitely believe that as more computing power becomes available, we will see manufacturing in all industries start to squeeze the blood out of every stone when it comes to the performance of any and all materials used to construct anything and everything. True artificial intelligence like Skynet or Cortana from Halo is not a concern any of us should have any time soon. Machine learning algorithms are a very far cry from actual independent thought.
Electronic triggers are present in pistols like Walther SSP-E. I think also Match Guns and Pardini as well. So maybe it’s not that there are no electronic triggers at all, it’s more about adoption for bigger calibers.
Re: Energy to power gizmos from thermal and kinetic energy while firing...there are a lot of troops, cops, and even non-LE civilians who go without shooting for years.
1:23:08 _best guess_ on my part is the LWRC M6 series. Same issue as the Colt M5. Of course why the Army cares is beyond me. The M# designations started back in the 30's, long before Colt didn't predict the future correctly.
21:00 just to give a more real-life example of what I think Ziga is talking about: Go Masters were shocked by moves that a supercomputer played simply because it was completely outside of what they had thought possible. Completely valid moves that were supposedly not-random but had implications on the game-state. So while I get what Ian is saying that all solutions have likely been discovered simply because there were a lot of bright guys who had to get around copyright restrictions, I also agree with Ziga that it shouldn't be dismissed outright: we simply do not know what we do not know.
If Ai can be used to improve the design of a suppressor, a concept that has been developed and iterated on, why wouldn’t Ai be just as able to improve on designs of all components? Right now Ai doesn’t “know” anything new, but it’s power to be used in iterative tests can help an engineer solve problems faster, CAD style programs didn’t necessarily make better designers, it made it easier to do hard things. Optimizing materials, improving ways to manufacture, designing to limit waste, improve reliability, or use a particular material in components could all be areas where Ai helps an engineer in any industry.
I think that the "Rounds expended per Kill" is very misleading. If you watch footage from the war in Ukraine there is a lot of guys just holding their rifle sideways above the trench and firing towards the enemy. It's not like aimed shots are missing because of the rifle type or caliber. I think that this type of suppressing fire is typical in any war and accounts for the vast majority of the rounds expended. Also, as far as I know, no one has ever been penetrated through-and-through by a baseball. So there's that regarding momentum as a measure of expected damage from an impact. 😉
Trying to scavenge energy from firearm recoil is not a great idea because it only would work when you are firing the weapon. Mag dumping to charge your battery might not always be a great idea.
For years now I've felt like Aluminum-cored projectiles are inevitable, at least for civilian use here in the US. It's the most common metal in Earth's crust, it's seemingly appropriately soft for projectile use (with a typical copper jacket, I assume), and it isn't on the US's stupid list of "illegal materials to produce pistol ammunition out of".
The problem is probably weight - to get a good ballistic coefficient you need a certain weight for a certain cross-section of the bullet That's why lead is so popular
@@PolenarTactical yeah, I suspect it'd work fine for some projectile shapes and just simply wouldn't for others. The SS198LF loading of 5.7x28 comes to mind - it's basically just an aluminum cylinder with a copper wrapper that gives it its conical front end. And it's stabilized by a 1:9 inch twist rate. So I think at least *some* spitzer projectiles would work fine in aluminum. They definitely wouldn't hold their energy as well over distance, but at least they'd be stabilized by pre-existing rifled barrels.
On the new rifle: most casualties aren't inflicted by small arms any ways. Mostly its used for supression, where plenty of ammo is more important. It is also extremely rare that you see the enemy, so the whole assisted scope thing, is problematic - for it is rare to see a whole person or even part of a person in combat. So can you do supressive fire then?
You would not believe the amount of troops in the military that cannot understand each others regional accents. Part of bootcamp is breaking down accents and developing a generic one that everyone else can understand. Most Southerners could not understand the family members from Northern MN that have the thicker accents and vice versa. I was 22 when I learned Boomhauer actually had dialogue in King of the Hill because I always thought he was just mumbling gibberish as a long running joke when I watched it as a kid. Also in reference to the Scandinavians in Minnesota, my great grandfather revealed on his deathbed he wasn't actually Swedish, but Finnish and immigrated to Sweden where he met his wife before immigrating again to MN.
1:35:30 By being the first to have a service rifle that can reliably penetrate, near-peer armies won't be motivated to adopt level 4 plates. If they're using level 3 plates or even no armor then the massive stockpiles of 5.56 stay relevant indefinitely just because the US can switch to the M7 at any time.
AI seems useful for ballistics research, but we sort of already tried every reasonable way to cycle a weapon _using the ammunition we have_ . Our self contained metallic cartridges produce gas pressure, which in turn makes the casing want to go reward when fired (i'm simplifying this), so we can either use the gas or the inertia from the case to cycle the weapon. What ai could do is maybe optimize those existing systems, maybe work with barrel harmonics and invent a conventional piston system that does disturb POI as much, but it'll still be a gas piston system. The laws of physics don't change just because we get access to a really smart computer.
"Printers" that work in metal becoming cheaper will change the gun industry more, most companies will go from selling guns to selling IP locked schematics/CADs. Any materials science advances AI comes up with will just get used in guns after the fact, those "Material" advances will not be born of an industry that has been "Innovatively/Intellectually Dead" for decades.
Finally someone (Ziga) addressing the elephant in the room: that the current "AI" is in fact just a virtual "intelligence". We are far from real machine self-learning and self-awareness.
AI is just that, Artificial Intelligence. Being self-aware is not part of the criteria for what makes an AI. That's only a thing in movies.
Machine learning is already here, and extremely common. In fact, if you were recommended this video, it was likely the result of machine learning.
Not really that much of a elephant in the room. There is going to be a fine line very soon between what is virtual "intelligence" and what is just intelligence. We may have created it, but that's a moot point if it keeps exponentially improving.
There is a reason the tech world is obsessed about "the singularity" because that's just how exponential growth curves work. Maybe human level or greater general intelligent AI is 100 years away, or 200 years, or it may be 5 years away. The honest truth is, even at top level research and universities, we don't know, it's allot of guess work and assumptions being made, there are many very qualified and very smart people that think we are closer to reaching that goal in 5 years than 100. People would have laughed in your face not to long ago about everyone having a super computer in their own home, now we have super computers in our pockets, hundreds of times more powerful than the entire computer system used in the Apollo moon missions. Even old science fiction didn't go that far in their predictions.
A point about exponential growth that even very smart people have a hard time grasping. If someone asked you what would you rather have, $500,000 or 1 penny today that doubles each day for 1 month, the vast majority of people would take the half million, yet starting from just one cent and investing it two-fold the next day, by the end of thirty days, it would grow exponentially - to a massive $5,368,709. Self improving AI has a potential for exponential growth, and allot can happen in a very very short period time. We also often forget that for thousands of years humanity went from 90% farmers and riding horses, to industrialized society, supersonic flight and traveling to outer space in the span of a single lifetime, and technological progression is still rapidly growing.
@@LN997-i8x I could argue that the criteria is incredibly low. But that's a philosophical view, I know. One could say that "if sentences" and "if programming" does not constitute real intelligence but then again, even humans are "programmed" through life by the same principles and you can fully understand one human being through "if matrix". And it's the same with abstract thinking. Abstract thinking is how we define in animals which is intelligent and which is not. Is the current AI really abstract thinking, if it has pre-programmed answers? Then again, is a human?
@@-Zevin- "elephant in the room" from a marketing perspective. On one end you have what the scientific community is doing and on the other you have "if algorithms" being marketed and sold as AI. And now everything is AI, from your coffee maker, Alexa and whatnot.
@@perunikavarnost6472 Yeah I can't disagree there.
Just noticed your name too Перýн, pretty cool, don't come across the Slavic pantheon very often. I had a axe of perun myself cast in bronze from a old archeological find, wore it on a necklace for ages.
I don't want my rifle or pistol to be sentient enough to tell me how my marksmanship sucks.
😆
Or how much
😂 That an easy fix, suck less!
Intelligent weapons in Dungeons & Dragons - "That swing sucks, pussy! My late master did better than that, even if he did get eaten by that dragon..."
AI would design a M-25 40 watt plasma rifle that fits perfectly into a skeletonized metal hand
40 wotts??
@@EFLO3D u wot M8?
Everybody needs a phased plasma rifle in the 40 watt range.
“Hey just what you see pal!”
45 long slide with laser sighting??
I'm an American doing a semester abroad in the Czech Republic and it's been a mindblowing experience. I've never been to Europe before and have been amazed by how different everything is. I plan to try to make a trip to Slovenia at some point.
must be nice to visit a majority white country where it doesn't look like crap everywhere huh?
I'm from Belgium and i can recommend Slovenia. They've had recently very bad floods check if you go any time soon if it would affect your travel.
@AlexLee-dc2vb if you are in CZ certainly stop by the store on Opletalova Street in Prague. Guys there will give you some recommendations what cool ranges to visit. :)
I see Gun Jesus I click.
Man of culture.
Amen - The Polenar Tactical Crew Rocks too!!
You’re not the only one
That's how I ended up here.
Noice
Seeing this podcast is giving me flashbacks already. It was a pleasure to compete with you guys at Lynx Brutality 2023. Can wait to see you guys again.
I think non-intelligent AI shouldn't be underestimated for firearms development:
"Take this design and reproduce it with less parts to manufacture", "take this blueprint and lower the pressure on the bolt face", "find a way to combine these two functions in single part with improved geometry "
Those are the type of prompts that would require very little intelligence, but would take months for a human engineer to crack
Really good thinking, even just these simple commands could do a lot. But them imagine a really advanced system that can make the optimal caliber/ammo/rifle combination
Caseless ammunition when?
This sounds less like AI and more like a very specific program similar to digital wind tunnel testing or scientific fluid dynamics programs. It's all dependent on the parameters that you put into it when you write the code, and that has to be VERY exact, to the point where it's more work to write the parameters than to do the job yourself. The program would need to know things like "yes, aluminum is light weight but also a bad choice for a barrel". It also has to know what you consider an "improvement". And how to weight these different demands. Is fewer parts more desirable than longevity? Than user safety? And then it has to take human ergonomics into account. No, that's where the user's face goes, you can't have the bolt moving into that area, silly AI!
If you really want to combine two functions into one part with improved geometry, just hire a Soviet engineer who works only with pen and paper (Sudaev's recoil spring guide rod which is also the extractor). Or if money's not an option, Christian von Koenigsegg (engine and liquid tanks are load bearing parts of the chassis). Both will do this better than any computer, and probably faster too.
If nothing else it would speed up the process if you tell it things like, "Smooth out the pixelated surfaces on this enlarged image." "Fill in the negative space."
I feel like Ian's point about that we have already tested so many systems and we are left with the ones that worked the best is a bit premature on what has been tested. It kind of draws the conclusion that what has been tested is essentailly every possibility and we are left with a handful of sytems that work well.
In reality I have the feeling it's more like the designs we have today is a fish bowl, and what we have tested is a swimming pool. But there are entire oceans outside of that that we haven't bothered with because the inter-disciplinary distance is too vast or complicated to justify the time/cost. And imagination isn't limitless, there are limits and good useful imagination is only bestowed on the few, unfortunately. What we're seeing from AI is a way to sort of bruteforce our way to a breakthrough by testing every possibility (instead of waiting for someone to connect the dots once in a blue moon), and knowing what it may look like. A proper AI would not need to draw borders between different disciplines, and will walk between them easily and utilize every tool and discovery humanity has amassed, and even come up with its own ones as Ziga mentioned, and focus it into a single goal.
That was a bit long winded but I hope my point comes across, AI is not limited like humanity is and the ideas and frameworks we build up for ouselves.
Bingo!
The designs that we have now are the ones that are logical to use because we know how they work and we were successful to mass produce them relatively cheaply. But this is far from all the possible systems - some that could be vastly better and more efficient, but we would need to look outside of the box.
Similar to what Elon Musk did with space travel - reusable rockets were "too expensive, too complicated and too expensive to make" - until the madman did it and revolutionized the space travel. Now it seems logical to us that reusing rockets is the way to go because it 100x to 1000x cheaper than building new rockets everytime. But until someone did it successfuly, everyone thought it was an impossible idea
So yes, that's one of the points i was trying to make, with tghe correct parameters and a defined goal, AI could probably come up with very interesting and efficient ideas.
electric weaponry is the future. the theoretical limits are SO far out, that laser and rail weapons (or coil) are possible even within what we can reach today! (far more interesting, once we break the fusion reactor technology barrier).
smart weapons will be the next step. tiny gyros correcting for you shaking etc. smart bullets that can correct themselves etc.
if you like, you can take a look at the alita manga for some hints of what is conceivable. one example are the "hornet bombs".
tiny drones the size of a giant hornet, able to seek out your vulnerable spots (like the neck) where the armor is the thinnest with enough explosive to just penetrate. no wasted detonation energy.
same for larger scale weapons. you have practically three limits.
- physical barrier, where penetrators are always stronger than shields at some point (due to weight and structural limits)
- energetic barrier, where lasers are probably the easiest thing to achieve, but basically every particle or wave than can be moved, can be used to cause a change in the target system
- metaphysical barrier, this is only hypothetical, but in theory you can use quantum linkage or quantum shift to break apart the links between atoms in your target or cause fluctuations in the reactor cores etc.
apart from that, the conventional ammunition will, at some point, reach its efficacy limit, due to "APS" countering.
shaped charges can only dig so deep, pressure spikes can only penetrate so deep.
etc. to break the limits of this ammunition, you will go towards the two barriers named above: by either going more physical (coil/rail guns, mass accelartors, orbital penetrators etc.) or more towards energy exchange (i.e plasma generators, tiny nukes or things like that).
addendum. flechettes are practically physical penetrators, just with a "flagrating" delivery method. once you can remove the energy waste and just accelerate the dart in a different way, they will come back.
though technically, "bomb" shaped projectiles will outperform flechettes and "conventional" shaped bullets (small, to medium) at some point. you see that in how most super long range bullets are shaped like an offset oval, where the back end is slightly tapered and the front mostly have a flat or rounded nose. (instead of this: ^ )
@@AkiSan0its like slap rounds, or 120mm discarding sabots but in a handgun or rifle formfactor. I think 3d printing will be the only way to make that type of bullet cheap enough for general use. Some kind of metal dart like object with a hard plastic sabot.
This would be really expensive but with every "bruteforced" iteration/design that seemed among best the testing firm or company will fabricate at least 10 prototypes per design(widening the chance of failure so that we don't make a huge mistake if we only have one prototype per design but is rejected on a production fluke) and put through numerous proving tests to work out lubrication and gas pressure issues(or possibly spring strength if it used that old piston idea as opposed to Stoner's AR gas impingement) and then after averaging out actual field test data select the empirically best one(a decision based on both shooting statistics but that is only half the other is realistic physical characteristics judgement.) A very technocratic Darwinian funnel where through the top we drop in potentially infinite but realistically still alot but a good real number of prototype designs and out of the bottom emerge only fittest from the beginning survive/adapt to become the fittest. And the best part is if this is done in the private sector it will be very efficient in the name of profit unlike government jobs which all service veterans like to chip in and say feels more like bare minimum at best or incompetent bureaucratic quota checklist at worst. I don't see the state taking up the testing role anyway very rarely do innovations come from the state rather mostly adopted from the private sector, but if they did all this would be extremely classified and more taxpayer expense.
One thing that no one brings up when talking about development of new firearm technology is powder. The current smokeless powder was a huge technological leap adding more range and cleaner burns so that semi automatic guns could be designed. If we could invent some new propellent that is more powerful and efficient than current powder it would change the design of firearms completely.
I work pretty deep in AI/machine learning (happy to prove this) and as a gun guy I actually have some interesting ideas about this. Would be super interested to hear Ian's takes on this especially.
Žiga and I must have similar reading interests. I have read that 45,000-50,000 rounds of US ammunition were expended to achieve one enemy killed in WWII and Vietnam. GWOT numbers are supposedly around 250,000.
Such high numbers are exactly why many countries hesitated to abandon bolt actions even in the 1940s. One fear was soldiers would shoot more ammo than a country could produce/supply if they were issued semi or full auto small arms.
There was even hesitance to the first single shot trapdoor rifles for the same reason. The British 1888 Lee-Metford had a detachable box magazine, this was highly controversial, and soldiers were only equipped with a single spare, and the primary mag was attached to the rifle by a bit of cord so you didn't lose it. They were worried soldiers would waste ammo, so really tried to drill into the men it was primarily for emergency use only, when you really need to fire and reload quick in a pinch, but not as standard operating procedure.
At 1:40 you're all talking about range and limitations of cartridges. With a couple of pals we did a very unscientific trial of 7.62x51 (147 German ball surp), 7.62x54R (STV/Norinco surp )and 7.62x39(STV/Norinco surp) at a 1200m range in Sicily. All the rifles were with iron sights, no scopes. Basically, we could make accurate shots on 80x60cm steel plates out to 700m with the x51 and x54R. After that the rounds went trans-sonic and accuracy fell dramatically. The x39 from a standard 16" Cugir AK74 wouldn't make hits over 500 with any accuracy. The other rifles used were an AIA No4 with 24" barrel and a Finnish Mosin M91/30 which is basically new (1944 Tikkasalo barrel), came from Finnish reserve stocks. Having carried the FN FAL and then the AUG in service I'd rather more ammo in .556 than a bigger round - more bullets, more hits....
You can extend the 7.62x51 up to 1000 meters with the correct ammo. Yeah, most AKs have a limit at 500 meters (again, depending on the round and altitude) as the bullet comes in transsonic ranfe and starts to irregularly deviate. I remember testing my AK at 480 meters with a new barrel, nice scope, good ammo - everything was perfect to make the hit. At 450m i was hitting the target but at 480 meters i was aiming at the center of the target but each shot hit differently around the target - to high, to low, 1 meter left, 2 meters right 😂 it was funny but also very educational to see.
There are special heavy bullet designs that are made to be stabilized when coming from supersonic to subsonic flight but those are quite expensive and as far as i know, only work in .308 and not 7.62x39...
@@PolenarTactical I run a Savage BR12 in .308 with a 29" barrel. Scoped with a 12-50x56 S&B PMII. I can make accurate shots out to 1200m with handloads using 180gn Sierra heads, Lapua cases and RS50 powder...as long as the shooting gods are with me... 1100m is no problem.
We had the same fun as you with the x51 and x54R at 800m, up, down, left, right...
"Sponsored by Schmeisser"... that one's not going away anytime soon. 😂
Great discussion on xm7. I personally believe that the Army plans on the combination of armor penetration and the new optic changing infantry tactics from “suppress and move” to “see and kill”. I’m interested to see if they are able to implement this.
Having watched the whole thing, you guys really made for a fun watch. Five guys comfortable with each other and knowledgeable about firearms. Just like a night of friends shooting the shit.
I think what Ziga is missing about 6.5 Creedmoor is that it burns barrels out way too fast. That’s fine for competition and snipers, but not infantry rifles or machine guns. And I say this as a big fan of 6.5 Creedmoor.
Oh, im completely aware of this - faster projectile, higher pressures = reduced barrel life
But look at the new M7 rifle, it's velocities and max pressure of the round - apparently that barrel can handle it without a problem. So if someone wanted to make a good 6.5 CM barrel that would last, they could. We have the technology but it probably doesnt make financial sense to produce such an expensive barrel for sport use because it's probably cheaper to replace the normal barrels more often
Does it, though? 556 is more "overbore" of a cartridge and 6.5 Swede and Japanese were commonly used in machine guns without major issues (apparently). The Army is also content with going with a much more potent 6.8x51.
"Motivational AI": "If you do not run faster I will taze you" 😂
A.I.: you don’t need guns anymore, so just throw them all away. Why do you need guns, A.I. will protect you.
Average Gun owner: 😳
About the issue of armor penetration, there was a level 5 plate tested that was far superior to XSAPI and the only round that penetrated it was 7.92 Mauser SMKH Tungsten rounds (Buffman test on Adept Collosus). Where 300 win mag swiss PAP and 7.62 NATO SLAP failed and custom 300 win mag. The difference is the WW2 tungsten rounds had a much larger core so mass makes a difference for armor penetration as well as velocity. The Germans duting WW2 used a powder (used in SMKH and v-rounds) that could bring greater power without the pressure being too high. The US and Soviet Union were never able to replicate it. The SMKH shoots a 194 grain bullet at 2900 fps which is wicked hot. 6.8 catridges would have too small of a core to do as much damage to level 5 as 8mm. Lethality over range is why 6.5 cartridges were considered insufficient in the World Wars. Honestly a modernized 8mm (even 8mm ball is supersonic past 1000m) or something similar would be way better than a 6.8 for destroying armor at longer ranges. A heavier cartridge would be better for machine guns and DMRs and then most infantry should stick with a lighter cartridge for closer range. Most soldiers won't even shoot at an enemy past 200m. An MG will be far more effective at range than even 10 riflemen.
The pint i think you are missing is that the large military application of the new cartridge doesn't aid the function of a platoon on fire and movement because the less quantity of ammo. There are situations where you have to spend ammo even if you have a precise optic for suppression and "stress". And the other and that at 400-600m, even with the optic, is not easy to spot people that try to hide and shoot at you.
Ian @Forgottenweapons @1:38:00 the other pro for the new rifle is barrier blind or terminal performance after going through cover. As well as eroding intermediate cover independent of belt fed support weapons. Like Ian has said the balance between issue rifle and GPMG/LMG has been contentious since intermediate caliber rifles Became standard issue.
jari is so freakin' awesome and he seems so calm & zivilised between you animals 😂 but i'm pretty sure he is the toughest dude on the table by a long shot
Jari is the GOAT
Been trying to coin the term "Kaptajn Suomi" for him for a while, you know, as in captain America. Feels fitting but doesn't seem to catch on
14:19 - Ian is looking for the terms "path dependence," "local optimum," and "global optimum."
Evolutionary trap is a real term too though.
This was a very enjoyable video. I hope you guys do more of these type of conversations
Thank you, that's the plan :)
1:08:08 the sabots for the flechettes go in random directions potentially injuring unintentional targets. muzzle blast was referred to as "massive", probably causing issues. flechettes themselves were sensitive to any sort of barrier or wind deflection. I just read up on this lol!
I always love how chaotic podcast gets when there are so many people around the table. Sometimes it gets a lil bit too chaotic but mostly it was a blast, just like hanging out with friends. The vibe is there. :)
Hahahaha 😂 That's Crazy Ian Mcollum's Forgotten Weapons Pissed Funny😅😂😆😄
Can't sleep. *sees 2hr polenar video* Still can't sleep but now gun talky noise in background
I click save and have them on demand. Talk shooty dudes - Mansa, talk.
Play Forgotten Weapons videos - sometimes when i watch his longer videos after dinner i fall asleep in front of my screen :D
His slow, calm voice is perfect for this and as an added bonus, i learn new stuff about guns
Gomna watch this tomorrow morning with some coffee. Cant wait
when it comes to xm7 there is always an option to switch the barrel back to 7.62, so if the new cartridge fails it can always be used as a 308 dmr
Or an actually useful 308 Battle Rifle.
Where I think AI will be relevant in firearms is sighting systems and heads up displays. A sight with passive, image recognition based range finder, ballistic computer, and possibly later threat priority indicator for example is clearly within the reach of even our current level of capability. At the very least it can be done poorly in a lab. But that is the first step and how things like night sigths started as well.
This is great. I need to check out the other episodes of this podcast.
@polenrartactical @1:48:00 I can say mortaring an X95 broke the charging handle and bent out the polymer guide for the handle. It broke the solid metal rod so maybe it was inevitable. The F2000 was not fun when I tried because the ergonomics were very different than AK/HK/AR
Another idea about AI drones: A quadcopter with a IR laser and thermal imaging.
You're on a mission. At night. Dual NODs equipped. You're in Ukraine. Looking for enemies in a forest line.
Your drone flies around, *looks* for targets via *Thermal* imaging - and once seen, it *points a laser at it/highlights it with IR light*
*You* would give yourself away shining IR lights, pointing lasers everywhere - but if the drone shines light at enemies/points at them with laser - it's useful.
You'd be equipped with a weak radio signal emitter, making you not a target for the drone (so it doesn't give you away).
(kind of like how current DJIs can "follow" the user based on the signal from the controller)
(signals give you away to enemy strategically, but once you turn it on within 200m of enemy positions it won't matter much)
There were complaints since Fallujah that 5.56 wasn't immediately lethal enough against certain targets. I think that perception of the troops is important. If the troops aren't confident in their equipment it has negative effects on the battlefield. Troops are less likely to charge the enemy of they don't think they can take them down reliably before they receive return fire.
I know 5.56 is fine against people, but I do enjoy .308's punch more, especially against non-human targets. I'd sooner trust the 240B to stop the VBIED than my poor SAW. With my SAW I'll have to aim at the driver's face, the 240B I can aim at the vehicle's engine if I can't reliably see the driver. I wonder how does the new cartridge perform against vehicles and cover?
Take the laugo alien. Red dot with an electronic eye that can take a picture of a target. Retrofit an electronic sear. Program the sight to fire when it matches the first picture while the trigger is pulled. Would need a button to reset the target memory. Accuracy would be determined by the first shot, but the followup shots would be super fast.
Man I love Jari. He's funny and so Finnish common sense. Intelligent, down to earth but with a sense of humor.
Best intro ever.
Regarding AI in small arms design, one can just look at algorithms like generative design to get a glimpse into the future of AI in small arms design.
The High Council gathers, what a wonderous day!
Missles already can predict where the plan will be when launched. We have no idea how deep it goes
Your actual prompt needs to be "how to win the war", "how to reduce suffering ASAP", "how to avoid war in the first place", etc. As soon as your prompt is _about guns_ you're basically making shooting people the goal. I think the best use for AI is out of the box solutions like the whole "intermediate calibres so it's a war of attrition against their hospitals, transportation, etc" thing (using an existing example because I kind of have to). Like Ian says, every possible _mechanism_ has kind of already been invented.
Funnily enough long recoil and blow-forward are actually good solutions to questions other than "how to make the best gun". That's why the latter is all over Evan Jones's "Future of Printed Rifles" slides, where the objective function isn't "make guns better" but "make guns easier (to build)".
"every possible mechanism has kind of already been invented" this is absolutely not true and just shows how our thinking is limited inside the bounds of our successful creations and not real physical limits
@@PolenarTactical Yeah, I guess so. The reason flechette theory (speed kills, moreso than mass) works for tanks but "not" for small arms is that people noticed rifle flechettes get blown off course by wind and rain at rifle engagement distances and then went "oh flechettes don't work then". I contend that flechettes make perfect sense in the PDW role.
A new fundamental machine, the rolamite, was discovered in like the 70s. That's as wild to me as a new Platonic solid being discovered. It has never been used in guns.
Nic Jenzen-Jones in his ARES report on telescoped rounds pointed out that the usual psi limit in weapons is actually due to the brass brazing to the bolt past that limit. So it isn't a real, _fundamental_ limitation.
Double-action triggers are never seen in long guns (except the 7.62 part of the Croatian grenade launcher), because there's no reason they're needed (in real steel), but I have pointed out elsewhere that sao sitting at full cock all of the time is a pretty bad match for printed plastic.
But except for the rolamite those are all already-invented fundaments that you might use because the objective function changed. When it comes to say operating systems, it makes perfect sense to me to say that there's a fixed number of energy sources (recoil, gas, external power, a few others) and a fixed number of ways to use each one, and all have already been discovered. Same thing goes for say feed systems.
The only truly new element I can see happening is electric propellant (see Integza - VERY different from either electric ignition or electric firing, which have always existed in autocannon). Then you could get closed-loop control of the burn rate, you could go subsonic based on a selector instead of a magazine change, etc.
I was in Texas. I drove four hours. Nothing changed.
Ziga, it's Texas. Four hours? You didn't even make it to the next county.
It does actually make me wonder, what kind of cursed Shenanigans can AI design lol
47:30 cortana, wake me when you need me
This was great!
Glad to hear :)
It doesn't matter if you can hit a man at 800 metres, if you can't see him to begin with he may as well not be there as long as he doesn't see you. We also could do something like a MK12ModH in 6mm ARC and get the same hit capabilities if you slap the XM157 optic on it.
It's good to have a marksman with an XM7, just like we used the M110/MK11, but even as we saw with those weapons people didn't like em. The MK11 led to the MK12 for instance.
The other issue is that a massive change in equipment will mandate a change in tactic, _but_ your tactic is only viable if the enemy allows it.
I thought the new M7 was a result of learning the lessons from GWOT especially Afghanistan where the engagement range was much farther than 50-200m and was from mountain slope to mountain slope? I may be mistaken.
An AI system should be able to model and analyze more iterations of a design in a day than an engineer could do in a year. That's where I think you will see it manifest in design. You could get some funky designs though as AI common sense.
Based and forgotten pilled
There are too many things which I wish to comment and most of I forget while listened it to the end...
About AI in gin industry. Computer can't think outside the box, it only operate it's algorithm in which was included existing parts. However it can help to iterate and fine tuning of already developed idea. Just, for example simulate different weight of some parts or search ways to lightened without causing structural weaknesses.
That's the difference between a computer that you think about now and real Artificial Intelligence - even now we have really good machine learning that takes certain things, analyses them and gives us completely new and unheard of solutions (for example medical prognossis, human genome,...)
Even with current machine learning and AI we only program the core of the system that then learns by itself and creates new code and new parts of itself. Look at the Chat GPT - even the people that made and trained it do not completely understand how and why it can give certain answers in a was it does
Going into the cartridge discussion on the new US rifle weapon system - the questions being asked here are incomplete, if not improper. Does a Level 4 plate change the field? No. Then defeating it is immaterial.
Moving from single-shot, limited-capacity, slow-reload rifles to detachable box magazines with high capacity and quick follow-up changed the way you could conduct warfare. Rather than engaging an individual combatant, it allowed you to lay suppressing fire to deny movement while mortars, field guns, artillery, and later air support did the work.
Having a Level 4 plate may defend vitals against rounds or shrapnel, but it does not allow the combatant movement in the face of incoming fire. Regardless of whether the opponent is firing 5.56 or not, the flow is still suppress with small arms and use fire support. The additional ammo capacity of a smaller, lighter round allows more intense suppression or more duration of suppression.
Rather than smart rifles, you would be better off investing in smart mortar launchers. A system that can integrate GPS coordinates and optically spotted or datalinked positions for more accurate, more rapid fire. Superior fire support is the proper answer to body armor, not a small arms race.
Flechette rifles would've been amazing if they actually worked. They're essentially the poor man's railgun.
I predict we will come closer and closer to flechettes over the years with development - this is because longer projectiles have higher ballistic coefficient, lighter projectiles can be fired at higher velocities which flattens the trajectory, reduces time to hit and can actually deliver more energy. Also lighter and longer projectiles will mean that soldiers will be able to carry more ammo
Imagine making a match dedicated to trench fighting. Trench Brutality
So, there is a MOSFET which operates the trigger.
You have set it to double-tap a particular shaped target on the timed course.
Your time runs out and you must stop, your trigger has stopped firing at those targets anyway and reverted to its default personal defence state.
But before you can Show Clear it recognises the face of the R.O.
I got zapped once in the stage and man it felt like some on kickt me in the ass while standing up
So you got the full experience :D
@@PolenarTactical @sgtmett6946 I hear they even had Hardcore mode, not pussy mode!
The current AI systems we have are very much AI. The "AI Effect" is a known tendency to dismiss any new AI advancement as "not real intelligence."
The technology is going places, of course, but the fruits of over half a century of AI research with agent theory, action planning, and more is all around us. Everyone keeps shuffling the goalposts around.
23:30 I for one, welcome our new AI overlords
29:46 We don't even have automatic doors that can do this
True, but it doesnt make any economical sense to integrated such technology into simple doors. But we already have the technology as it can derive from "simple" machine learning of multiple soldiers
New podcast, good shit 👍🏻
How does this video only have 30k views? Its so fun!
To your thoughts on the AI allowing you to shoot when you are on target: What i can imagine is the guns allowed for sport shooters only being capable of shooting at normed targets.
Theres those cheap shake to charge flashlights that moves a magnet thru a coil to charge tge internal battery. There are watches that motion charges the watch.
AI in weapons (in my dumb opinion), won't be in design, but in operation.
Imagine a weapon that monitors target position, movement, shooter biometrics, etc and says to the shooter "don't take the shot now. Your heart rate is way up and there's unknown walking in the shot from the right"
Yup. We should want to automate things that can be automated.
41:00 powered rails were one of the selling points for the textron suggestion for NGSW projects. yes lots of energy is released but how do you capture that? not too many ways to do so. but you're onto something.
Instead of a buffer spring you could use a magnetic shocks, thise magnetic shocks could be involved in electricity generator. Thermo-electric generation included.
1:34:22 To Jordan’s point, in a protracted conflict like we see in Ukraine, supply shortages means artillery rounds may not be available, is being rationed, or local artillery is worried about being counter-batteried. Additionally, contested airspace could mean CAS is unavailable. In that context, an M250 with optics sounds like a pretty decent weapon.
There's a bunch of Finns in Michigan's upper peninsula.
34:26 Pressing F1 to enter the BIOS
how does czech beer always ends up in your videos? :D
because Czechs make good beer :)
What AI designs is based on the programmed information. If you want an indestructable rifle that's inexpensive and reliable, you'll get an AK.
The machine learning "AI" that we see today will affect the manufacturing side in the way Ian described it with iterating suppressor designs. Medical research is already inching towards using machine learning to help scientists trial new drugs by iterating through different chemical compositions. This could happen with bullet design as well. Machine learning AI will be able to iterate through millions of bullet shape, material construction, and twist rate designs, simulating the ballistic performance of a round, or simulating the transfer of energy as the round hits a soft or hard target. This will lead to new bullet standards. Machine learning could also iterate through countless combinations of case dimensions, gunpowder chemical compositions, or even the size and shape of a grain of powder to find the most optimal configuration for perfect powder burn. I definitely believe that as more computing power becomes available, we will see manufacturing in all industries start to squeeze the blood out of every stone when it comes to the performance of any and all materials used to construct anything and everything.
True artificial intelligence like Skynet or Cortana from Halo is not a concern any of us should have any time soon. Machine learning algorithms are a very far cry from actual independent thought.
Electronic triggers are present in pistols like Walther SSP-E. I think also Match Guns and Pardini as well. So maybe it’s not that there are no electronic triggers at all, it’s more about adoption for bigger calibers.
Re: Energy to power gizmos from thermal and kinetic energy while firing...there are a lot of troops, cops, and even non-LE civilians who go without shooting for years.
1:23:08 _best guess_ on my part is the LWRC M6 series. Same issue as the Colt M5. Of course why the Army cares is beyond me. The M# designations started back in the 30's, long before Colt didn't predict the future correctly.
Im sure there are AGI already. At least the beginning of it. Just not anywhere the public can access it.
21:00 just to give a more real-life example of what I think Ziga is talking about: Go Masters were shocked by moves that a supercomputer played simply because it was completely outside of what they had thought possible.
Completely valid moves that were supposedly not-random but had implications on the game-state. So while I get what Ian is saying that all solutions have likely been discovered simply because there were a lot of bright guys who had to get around copyright restrictions, I also agree with Ziga that it shouldn't be dismissed outright: we simply do not know what we do not know.
8:00 what an insightful question
"Old West" have German-oriented attractions because Winnetou xD
More ammo is ALWAYS a better thing.
If Ai can be used to improve the design of a suppressor, a concept that has been developed and iterated on, why wouldn’t Ai be just as able to improve on designs of all components? Right now Ai doesn’t “know” anything new, but it’s power to be used in iterative tests can help an engineer solve problems faster, CAD style programs didn’t necessarily make better designers, it made it easier to do hard things. Optimizing materials, improving ways to manufacture, designing to limit waste, improve reliability, or use a particular material in components could all be areas where Ai helps an engineer in any industry.
Slide Rules take a little longer.
There's the potential. The thing with suppressors is that internal structure can now be refined due to 3D printing with sturdy materials.
I think that the "Rounds expended per Kill" is very misleading. If you watch footage from the war in Ukraine there is a lot of guys just holding their rifle sideways above the trench and firing towards the enemy. It's not like aimed shots are missing because of the rifle type or caliber. I think that this type of suppressing fire is typical in any war and accounts for the vast majority of the rounds expended. Also, as far as I know, no one has ever been penetrated through-and-through by a baseball. So there's that regarding momentum as a measure of expected damage from an impact. 😉
The perfecting a suboptimal design thing is kinda like Dawkins Mt. Improbable analogy.
Trying to scavenge energy from firearm recoil is not a great idea because it only would work when you are firing the weapon. Mag dumping to charge your battery might not always be a great idea.
From 5 days later: I had that very thought. Most troops don't fire a single shot aside from weapons qualifications.
There are companies working on harnessing the buffer in AR's to generate electricity during recoil.
Estonian cider, heck yeah
For years now I've felt like Aluminum-cored projectiles are inevitable, at least for civilian use here in the US. It's the most common metal in Earth's crust, it's seemingly appropriately soft for projectile use (with a typical copper jacket, I assume), and it isn't on the US's stupid list of "illegal materials to produce pistol ammunition out of".
The problem is probably weight - to get a good ballistic coefficient you need a certain weight for a certain cross-section of the bullet
That's why lead is so popular
@@PolenarTactical yeah, I suspect it'd work fine for some projectile shapes and just simply wouldn't for others. The SS198LF loading of 5.7x28 comes to mind - it's basically just an aluminum cylinder with a copper wrapper that gives it its conical front end. And it's stabilized by a 1:9 inch twist rate. So I think at least *some* spitzer projectiles would work fine in aluminum. They definitely wouldn't hold their energy as well over distance, but at least they'd be stabilized by pre-existing rifled barrels.
On the new rifle: most casualties aren't inflicted by small arms any ways.
Mostly its used for supression, where plenty of ammo is more important.
It is also extremely rare that you see the enemy, so the whole assisted scope thing, is problematic - for it is rare to see a whole person or even part of a person in combat. So can you do supressive fire then?
Somebody tell Jari I love his hair, no seriously, nice hair dude.
You would not believe the amount of troops in the military that cannot understand each others regional accents. Part of bootcamp is breaking down accents and developing a generic one that everyone else can understand. Most Southerners could not understand the family members from Northern MN that have the thicker accents and vice versa. I was 22 when I learned Boomhauer actually had dialogue in King of the Hill because I always thought he was just mumbling gibberish as a long running joke when I watched it as a kid. Also in reference to the Scandinavians in Minnesota, my great grandfather revealed on his deathbed he wasn't actually Swedish, but Finnish and immigrated to Sweden where he met his wife before immigrating again to MN.
Incase none of you knew, people that are top tier in the situation are actually really good and clever people who see clearly. I'd not stress so much.
Staropramen is a good choice of beer - what's the other one?
Union is a Slovenian beer and the Somerby is a cider
1:35:30
By being the first to have a service rifle that can reliably penetrate, near-peer armies won't be motivated to adopt level 4 plates. If they're using level 3 plates or even no armor then the massive stockpiles of 5.56 stay relevant indefinitely just because the US can switch to the M7 at any time.
AI seems useful for ballistics research, but we sort of already tried every reasonable way to cycle a weapon _using the ammunition we have_ . Our self contained metallic cartridges produce gas pressure, which in turn makes the casing want to go reward when fired (i'm simplifying this), so we can either use the gas or the inertia from the case to cycle the weapon. What ai could do is maybe optimize those existing systems, maybe work with barrel harmonics and invent a conventional piston system that does disturb POI as much, but it'll still be a gas piston system. The laws of physics don't change just because we get access to a really smart computer.
Cider? I thought you guys were whiskey drinkers!
AI is great if you've ever wanted to hear Al Jolson sing Du Hast by Rammstein.
I absolutely love gun jesus
There have been very few innovative small arms designed after 1950. My thought is there is no incentive to create new firearms.
@37:22 i already could build a prototype like that.
"Printers" that work in metal becoming cheaper will change the gun industry more, most companies will go from selling guns to selling IP locked schematics/CADs. Any materials science advances AI comes up with will just get used in guns after the fact, those "Material" advances will not be born of an industry that has been "Innovatively/Intellectually Dead" for decades.