Posts Tagged ‘AI’

Kill Command is a 2016 film written and directed by Steven Gomez. The film’s premise [spoiler alert!] is a marine unit being sent to a remote island for an unscheduled training mission. They arrive without the typical greeting communications by human support staff, but they continue on with their mission and destroy the mostly oblivious enemy robots as expected. What they don’t expect is those gullible enemy robots are just a ploy used by a next-gen, mechanized combat robot referred to as a S.A.R. (Study Analyze Reprogram) unit and its contingent of more advanced robot foot soldiers to observe the marines, learn from them, and then apply their own tactics against them. This unexpected, second conflict unfolds with the marines gradually realizing they’ve been brought here under questionable circumstances and their survival is very much in question. The marines evolve from surprise, to dismay, and finally to grim concentration as they manage their retreat as best they can. The machines have them outgunned and outsmarted, and the humans know it.

The marines use some questionable military tactics. For the first third of the move or so, half the team neglects to even wear helmets. Under fire, they stand around or run upright—real soldiers would be hugging the ground, prone or crawling to keep as low a profile as possible. There’s a fair bit of one-handed rifle firing with dubiously high accuracy. The marines neglected to bring any heavy weapons, other than a mortar which is quickly lost. Unrealistic aspects apply to the robot side as well. For example, flying recon drones continually fly up to within an arms-reach of the marines. First of all, this makes their surveillance incredibly obvious when it doesn’t need to be—with today’s optics, a target could be visualized from kilometers away (not to mention what military satellites might be able to see from orbit). Even in a forest, where much of the action takes place, drones could detect and monitor targets from tens to hundreds of meters away and remain cryptic while doing so. Why reveal their presence and give up the element of surprise? In the film, the impression is the machines may not care about surprise—they don’t need it, but then why bother with this conflict at all? I suspect they come so close more for dramatic effect than anything else.

Only the white characters survive. The marines start out as a somewhat diverse group with two black soldiers out of six. That diversity is whittled down until only the white characters remain. While at least one of the two black soldiers dies a heroic death, I have to wonder about the plot choices and how they will be received by various film-goers.

The acting is adequate, but no one really stands out. The characters are all pretty thin—we never learn much about anyone. No soul-searching conversations while under fire here. No one asks any philosophical questions about the nature of artificial vs. biological beings, or how the murderous robotic soldier became self-aware or why it decided to study human combat in a live-fire, kill-all-humans format. The marines accept their fate with a surprising stoicism and not a single “game over, man!” (that could have been a nice homage to Bill Paxton, but oh well).

The film has a definite Terminator-esque  feel but without the unique appeal Arnold brought to his robot villain. The S.A.R. villain generates sufficient menace but without any memorable aspect (no interesting personality like HAL from 2001: A Space Odyssey). The CGI effects are well done and almost seamless, yet lacking creativity. And in these days of CGI mastery, good isn’t good enough. To be fair, this could have been an intentional choice to remain faithful to the combat robot designs already being developed by today’s militaries.

In my opinion, the robots underperformed. In future conflicts involving ground robots (“foot soldiers” which I suspect we’ll see deployed in the near future), they would be the perfect snipers—no breath control necessary, no body movement to still, unblinking eyes that can see beyond the spectrum of visible light. In the film, the robot grunts (mostly four-legged, heavy machine guns) miss a lot, don’t fire at a rapid rate, and in some cases are easily destroyed by single shots from the marines.

Despite these quibbles, the film offers a good glimpse of the future of mechanized war, but it’s unclear what role humans soldiers will play going forward. How will imperfect humans participate in future warfare without being quickly exterminated? It’s not an easy question to answer—will humans be able to fight at all? Or will they be confined to a bunker somewhere waiting helplessly for the cold calculus of machine vs. machine to play out on the battlefield and seal their fate as victors or victims? I’ve blogged about this topic before and took it even further in a short story, Lonely, Lonely. My guess is future human soldiers will only survive by working intimately with their own robot protectors (I’m exploring this idea further in a new series of far-future novels). Yet in reality, I suspect we’ll see early answers to questions like these soon enough. Just like the marines in Kill Command, we may not like what we see.


What is the trajectory of warfare? Using science fiction examples, we might assume the soldier of tomorrow will be fighting in a powered, exoskeleton suit like Tom Cruise in Edge of Tomorrow. Edge of TomorrowOr perhaps piloting a giant, human-shaped mecha robot like those featured in many a manga/anime series and most recently in the film Pacific Rim. What about humans battling in space à la Star Wars’ X-wing fighters and Battlestar Galactica’s Vipers?

All these are unlikely to become a reality. Soldiers fighting in exosuits or inside giant mecha would be too unstable. Human pilots would be too slow. All of them would be too vulnerable to remain on the battlefield.

  • Stability: A two-legged, armored soldier or mecha  doesn’t make much sense from a stability standpoint. Vehicles and robots with four legs and/or wheels or tracks would be much more stable, allowing them to move more quickly over rough terrain.
  • Physical speed: Even with an exosuit, human soldierBigDog_Snows are unlikely to be able to move as quickly as future ground robots. Typical human walking speeds of 3-6 mph are already being approached by robots such as BigDog which can already move at up to 4 mph. Factor in physical limitations such as exhaustion and muscle fatigue, and a biological system will inevitably fall behind.
  • Vulnerability: The human body can’t take extreme accelerations. Even the best-protected organ, the brain, suffers from the relatively benign impacts encountered by boxers and football players (American football). IED victims suffer similar head trauma. Human pilots can take extreme g-forces (10’s to 100’s of g’s) for only very brief periods (seconds to minutes) before losing consciousness or suffering more severe injury. Their ability to survive impacts or endure vehicle acceleration will never compare to that of a machine (such as the drone fighter in the 2005 film Stealth).

Putting aside human limitations, war machines are more likely to decrease in size and cost than they are to get bigger.pacific_rim_banner-wide

“Defense Department agencies are researching and developing relatively inexpensive “swarm” systems, which humans could supervise during operations. ‘These efforts hint at the next paradigm shift in warfare — from fighting as a network of a few, expensive platforms as we do today, to in the future fighting as a swarm of many low-cost assets that can coordinate their actions,’ Scharre said [Paul Scharre, director of the 20YY Future of Warfare Initiative at CNAS].”1

Regardless of size or design, will humans operate these new attack machines? They won’t. Humans would be the weak link in command and control.

“ ‘Humans will not be able to match the capabilities of autonomous systems when it comes to certain types of operations such as missile defense or cybersecurity,’ Work said [Deputy Secretary of Defense Robert Work]. ‘When you’re under attack, especially at machine speeds, we want to have a machine that can protect us. … You cannot have a human operator, operating at a human speed, fighting back against a determined cyber attack. You’re going to have to have a learning machine that does that.’ ”1

Today, drone operators represent a small fraction of military combatants, but I suspect that fraction will gro800px-MQ-9_Reaper_UAVw and eventually eclipse the number of soldiers on the battlefield. Today’s drones and guided munitions require satellite-based communication.  Assuming enemy states will disrupt those communications, the potential for disruption will push for more drone autonomy and less human decision-making. Eventually, autonomous drones will command and coordinate other automata to carry out missions, making their own operational decisions according to previously established priorities, decision-making algorithms, and rules of engagement.

As autonomous drones become the prevalent combatants other aspects of warfare may change:

  • Increase in risk tolerance and aggressiveness as human soldiers are no longer in harm’s way.
  • Decrease in cost per combatant as drones become cheaper per Moore’s Law (cheaper computer components, materials, fabrication processes) vs. the perpetual rise in costs to train human soldiers, feed and house them, and heal and repatriate them once combat is over.
  • Military power will depend less on the ability to recruit and train citizens for military service and more than ever on the economic base, industrial capacity, and R&D to outproduce and outperform the enemy’s drone fleet.
  • Battles will be drone fleet against drone fleet.

Where does it go from there? Following this trajectory, military escalation will push towards greater AI, more machine autonomy, while squishy, vulnerable humans are relegated to defensible areas. But how well can those areas be defended?

Why use a human sniper when you can field a rifle placed by a unmanned aerial vehicle (UAV) with a video aiming system controlled by a marksman (or anyone, really) thousands of miles away, puts just needs to put the cross-hairs on target and make the shot flawlessly—no rifle vibrations from an unsteady hand, breathing, moving, trigger pull—no need for a trigger at all. And why bother aiming the gun at all once self-guiding bullets are worked out? Mount a rifle on a UAV and you have a perfect sniper/assassin without the aggressor’s life being put at risk.

Go one step further down the road of miniaturization, and a future terrorist may be able to 3-D print a thousand UAV mini-mines, each with just enough explosive to murder an adult human by rupturing the victim’s carotid or femoral arteries or some other vulnerable point (eye, ear, etc.). The book Robopocalypse did a great job of imagining how machines themselves, once sentient, could re-engineer themselves to exploit the vulnerabilities of the human form: walking, heat-seeking mines for example. However, I doubt we’ll have to wait for the Singularity for this to happen. Militaries and terrorists will be more than happy to design these weapons much sooner than that. When/if the Singularity happens, it may be a moot point for the humans inhabiting that bleak future.

As weapons become miniaturized, autonomous, and cheap, how can society protect itself? Human-scale weapons won’t be necessary to kill a human target, nor could an exosuit defend against micro-mines. Only a suit of modern-day armor or perhaps a portable, rapidly repeating EMP (if one could be constructed with enough charge using a battery light enough to carry everywhere one goes) might be a plausible defense. Or perhaps everyone will sport their own phalanx of defensive micro-drones as in The Diamond Age by Neal Stephenson. Even with these potential defenses in place, if your drones are fighting enemy drones “face-to-face,” you have, in all likelihood, already lost.

All of this points to violence increasingly performed by machines. Human soldiers, behind the front lines, will be safely ensconced in bunkers of concrete and steel. But once terrorists embrace these coming technologies, how long will it be before we civilians go into the bunker as well? And when will we ever come out? Warfare will become a video game we play in our living rooms for very high stakes.

Taking it to the extreme, in my short story, Lonely, Lonely, humans have merged with and essentially become self-contained war machines: each person inside their own customized, all-in-one bunker, weapons factory, and command and control center. Only the most vicious of each generation survive until they have evolved to live permanently entombed in armored, metal shells, their bodies vestigial and their brains expanded out of their skulls to interface with the computerized components they control.

Given the difficulty of predicting the future, I hope I’m wrong and this is a future we avoid. I for one don’t look forward to life in an underground bunker, but maybe I’m being too pessimistic. We can always pretend we’re back on the surface under a blue sky—VR is about to break out, and maybe that’s all we’ll really need.

Do you see a different future ahead of us, or maybe a technology I didn’t consider? I’d love to hear your thoughts in the comments.



1Pentagon Seeks Smarter Machines for Future Combat, National Defense, March 2016, by Jon Harper.

Reaper image source: USAF Photographic Archives

BigDog image: