6 ms·
An AI defeated human fighter pilots in an air combat simulator (2016)
- squarefoot 9y agoA side effect of employing AI to pilot war aircrafts, like with smaller drones, is that there won't be families in grief for the loss of their kid pilot on the attacking side, so that warmongering politicians will get public approval of campaigns a lot easier.
- vmarsy 9y ago"just defeated" but should have a 2016 tag. Paper link for those interested : https://www.omicsgroup.org/journals/genetic-fuzzy-based-artificial-intelligence-for-unmanned-combat-aerialvehicle-control-in-simulated-air-combat-missions-2167-0374-1000144.pdf https://www.omicsgroup.org/journals/genetic-fuzzy-based-arti... (from a June 2016 press article : http://magazine.uc.edu/editors_picks/recent_features/alpha.html http://magazine.uc.edu/editors_picks/recent_features/alpha.h... )
- gtirloni 9y agoPrevious discussion: https://news.ycombinator.com/item?id=11993366 https://news.ycombinator.com/item?id=11993366
- coob 9y agoThe real kicker is that an AI controlled aircraft wouldn't be as limited in manoeuvrability as one that has to sustain a human.
- sneak 9y agoI feel like Terminator 2 should be required watching for these devs, no?
- Jach 9y agoNot really, because if you ignore the rest of the series T2 ends with humans winning, with the cautionary moral of 'no fate but what we make'. They need to watch something where the machines quickly go from seeming under control to dominating and wiping out humanity with no notion of a 'resistance'. Is there something in the Outer Limits/Twilight Zone archives for that? The closest thing I can think of right now is Stargate Universe's self replicating ships that unlike other replicators in the series actually won and couldn't be stopped, forcing the crew to hope they could make it to the next galaxy and escape (but if the bots are already there, they're screwed).
- dugditches 9y ago"Second Variety"(short story, worth a quick read) by Philip K. Dick was adapted into a film called 'Screamers'. https://www.gutenberg.org/files/32032/32032-h/32032-h.htm https://www.gutenberg.org/files/32032/32032-h/32032-h.htm https://en.wikipedia.org/wiki/Screamers_(1995_film) https://en.wikipedia.org/wiki/Screamers_(1995_film)
- antod 9y ago> Not really, because if you ignore the rest of the series T2 ends with humans winning Coz a pointless and massively destructive war is OK as long as we end up winning in the end?
- Jach 9y agoNot exactly OK, but it along with most everything else instills the spirit of muddling through to victory even with big screw ups. Better to fear extinction than mere war.
- sneak 9y agoColossus: The Forbin Project. One of my favorite films.
- jlebrech 9y agoall they need for the robot uprising to happen is to code the AI to protect humanity from itself.
- k_sze 9y agoNow we just need to develop the X-9 Ghost hardware in real life.
- Nition 9y agoThe article doesn't say anything about what data the AI had. Did it have full access to the simulation, or did it have to "see" the opponent's plane, the terrain etc a bit more like it would in real life? That could make a big difference. It's easy to make a self-driving car for instance if you can feed it the position of every other car in the world, with their movement data, and all other obstacles.
- obstinate 9y agoIt seems very unlikely that they gave the AI global awareness. What would a test like that prove? The people who are working on military software are not idiots. The article also refers to ALPHA being able to defeat the human pilot with impeded sensors, which implies that it was being fed information from simulated sensor systems.
- StavrosK 9y agoIt would prove that an AI could fly, maneuver, target, shoot and take down enemy aircraft, with good enough sensors.
- huangc10 9y ago#OldNews, this article is from 2016. > What sets ALPHA apart is the genetic fuzzy tree decision system that can calculate an opponent’s movements or strategies 250 times faster than you can blink. That's just not fair. One more comment, can this thing (ALPHA) land an Airbus A320 on the Hudson?
- tempestn 9y agoNot yet, but I can't imagine a future where AIs don't eventually control takeoff, landing, and emergency situations.
- phire 9y agoWater landings are a documented procedure for planes, there is a checklist (which Sully had practiced in simulators). Assuming an AI like this was given 100% control over a commercial plane, it would have been programmed with such a checklist. It would have also been tested on this ability in a simulator (just like regular pilots). The AI would have easily worked out in an emergency that it couldn't reach the runway, or any alternative runway and not even bothered trying. The only question is, would the AI have decided that the Hudson was a valid surface for a water landing attempt? And that really depends on the database programmed into the plane. It can't really see the river, so the database would need to have the river marked as water, at the correct elevation. It also can't really see the bridges (maybe if it had LIDAR), so the bridges would have to be included in the database too, so it could pick a good stretch to try and land it. Oh and boats too... either it would have to dynamically spot them, or get lucky. But if the database correctly reflected reality, then it would see the opportunity, try to execute it and potentially succeed.
- autokad 9y agoAIs are still struggling with 'is that white space a tractor trailer or sky? eh just drive through it' with cars. they are immensely far off from aerial combat and emergency maneuvers in aircraft. and the whole point of emergencies is that they are a bit unexpected and the situations tend to be rather specific. I'm not saying it wont eventually happen, but no: they unequivocally cannot fly our aircraft in takeoff/landings, emergency situations, etc.
- NicoJuicy 9y agoFrom reading: Amazing that the AI was running on a Raspberry PI... I'm amusing image processing happens on a "low resolution". Also, this seems to be interesting: "They tackled the problem using language-based control (vs. numeric based) and using what’s called a “Genetic Fuzzy Tree” (GFT) system, a subtype of what’s known as fuzzy logic algorithms. States UC’s Cohen, “Genetic fuzzy systems have been shown to have high performance, and a problem with four or five inputs can be solved handily. However, boost that to a hundred inputs, and no computing system on planet Earth could currently solve the processing challenge involved – unless that challenge and all those inputs are broken down into a cascade of sub decisions.” That’s where the Genetic Fuzzy Tree system and Cohen and Ernest’s years’ worth of work come in." "The training happend on a 500$ commercially available computer" "It uses linguistic input instead of numeric input" Can i conclude that ( after reading the article), that the training matches "situations"/"variables" to words. Eg. The enemy is far, close. The friend is behind me/in front of me/above me and then i go up/down/do a roll. Multiple generations later, some linguistic variables are left away, so only the necessary ones remain. From the comments on Reddit ( https://www.reddit.com/r/CredibleDefense/comments/4q8h8z/paper_on_ai_in_air_combat_genetic_fuzzy_based/ https://www.reddit.com/r/CredibleDefense/comments/4q8h8z/pap... ) "This is not another opinion piece vaguely arguing how AI will "one day" replace pilots. The authors seem very cognizant of all the challenges and complexities. They clearly limit and define the scope of their work." "This "research paper" is a commercial. The expert fighter pilot is an employee of the company and his job is to promote it." None the less, gonna read on some Fussy Logic :p - https://www.tutorialspoint.com/artificial_intelligence/artificial_intelligence_fuzzy_logic_systems.htm https://www.tutorialspoint.com/artificial_intelligence/artif... Edit: Link to previous HN discussion -- https://news.ycombinator.com/item?id=11993366 https://news.ycombinator.com/item?id=11993366
- AstralStorm 9y agoThey main upside of this system is that we can learn from it as well. (Though it does not explain they why well, it does explain the how well enough that we can figure out why it takes such an approach.)
- terravion 9y agoLove this! My least favorite part of sci-fi is that the heroes are always buzzing around in some advanced fighter aircraft. Crap, they might as well ride around on sci-fi horse and joust like knights. The era of the fighter ace is as gone as the age of horse cavalry--it would be great if the Pentagon figured this out before squandering literally $1T (not an exaggeration) on the Joint Strike Fighter.
- pavement 9y agoOkay, so aerial combat in the form of interceptor dog fights is an anachronism. Big deal. It's still worthwhile to develop high-performance piloted airframes, since people are still actors in any conflict, and will need to operate within the theater of combat. Even if primary operations are handled by unmanned systems, redundancy and fallback systems are needed. Not just troop transport from A to B. Well-armed vehicles capable of deploying weapons will still need to be piloted and maintained. Police still use horses. The military will still find utility in piloted warbirds.
- MarkPNeyer 9y agopolice still use horses, but they don't spend a trillion dollars on a new laserhorse
- lgas 9y agoIs there a petition somewhere I can sign to support the acquisition of a new laserhorse?
- stcredzero 9y agoI don't know of laserhorse. But someone already made a "LaserKlan." http://www.huffingtonpost.com/2014/02/28/colbert-laser-klan-viacom_n_4873687.html http://www.huffingtonpost.com/2014/02/28/colbert-laser-klan-...
- stouset 9y agoMost of that money was spent before drones were considered well-tested and viable. Would you rather we simply throw away the $1tn already spent and start over?
- brilliantcode 9y agoThis is like that movie where US pilots have to battle a super advanced AI stealth drone. AI really should have no place in military but it's going to happen. Some spineless fucks are going to spread their legs for a 5 star general to sell their company. The defence industry is in the business of killing humans and you play a direct role by selling your knowledge and skills to make it more efficient. Fuck that shit.
- landryraccoon 9y agoI don't think greed or spinelessness have anything to do with it. AI is similar to nuclear weapons : Any country that doesn't have military AI will have a huge disadvantage against a country that does. Every country rationally believes that AI will improve their military chances, and every country rationally believes that every other country desires AI for their military as well. China and Russia will certainly build AI drones as soon as they possibly can. The only thing that happens if western countries don't build them is that they are at a significant disadvantage.
- Jach 9y agoThis is also true of chemical and biological weapons, and weaponizing outer space, but countries have managed pretty well at not going into an arms race in those areas; even nuclear weapons have been reasonably well kept in check. But these mutual agreements to not pursue certain lines of military technology (at least for offensive purposes) really need to start from the strongest military agreeing to not do it and allow monitoring/audits to get other countries on board. The US doesn't seem all that interested in that with AI, or other scary technologies on the horizon. (Without even getting into the issue that AI is trickier than something like a chemical weapon, countries would have a field day on what actually constitutes AI vs normal digital automation...)
- landryraccoon 9y ago> This is also true of chemical and biological weapons I disagree actually. Nuclear weapons render chemical and biological weapons moot, and no major superpower is willing to give up nukes. The United States has no need for chemical and biological weapons because it's keeping it's nukes. I think that AI will present a comparable advantage. Also, AI does not (yet) create the sense of dread in the popular imagination that nukes did during the cold war. Incinerating an entire city in a few seconds is terrifying on an existential level to any human. Having computers control drones and tanks seems.. almost boringly inevitable, by comparison. I doubt there will be unified political will to prevent it. Also, I think the international climate has radically changed. Agreements like TPP, which the powers that be supported, were not able to pass. In the era of Trump I highly doubt there is enough global trust for Russia or China to agree not to research militarized AI. Basically I don't see a single major power agreeing not to develop militarized AI in the near future.
- shusson 9y agoWhy spend resources developing AI for a plane designed for Humans? I'd imagine that AI on cheap drones/missiles vs traditional human fighters would be a more interesting simulation.
- rbanffy 9y agoA good reason to let it control normally maned craft is because, since it and its peers will have a high kill rate, no human will want to fly a combat mission against them. There are a couple trillion dollars already spent in meatware-friendly fighters and it'd be a shame to mothball all that.
- davidwihl 9y agoIt would be a greater shame to send hundreds of well trained pilots in expensive aircraft only to get heavily defeated by cheaper and more flexible AI flown craft.
- rbanffy 9y agoThat's why converting those expensive aircraft to AI drones is a good idea. An intermediary step could be a hybrid system where the pilot can engage the AI to reduce their workload and work in tandem with the machine.
- graycat 9y agoFor some of that, there's some old and quite solid math -- differential game theory. See, say, Rufus Isaacs, long a professor at Johns Hopkins and who IIRC did his first work at a US Navy lab and see also Avner Friedman.
- chimen 9y agoOn top of that, it will make things much cheaper cutting the need to carry an extra 90~kg (the pilot), a huge interface with commands that must be operated by the pilot and a cockpit to carry him. Not to mention the human errors, and efforts invested in training them. Humans will most likely be used in "decision-making" roles.
- FrozenVoid 9y agoCurrent fighter planes are limited in their design to accommodate a slow-thinking, living human being. A plane can't reach certain acceleration, has to provide life support, space and computing power to cockpit instead of just functional components and armaments. When AI is polished enough, fighter planes will go the way of horse and carriage.