Remember meForgot password?
    Log in with Twitter

article imageOp-Ed: DARPA’s Squad X A.I. at Grunt level — A new kind of war

By Paul Wallis     Jul 13, 2019 in Technology
Washington - Think for about two seconds on the subject of infantry using A.I. in a firefight, or as tactical data in real time. Interesting, it definitely is. Good idea? Better than none, would be the obvious answer. DARPA's Squad X is a new threshold.
The basic idea of Squad X is to deliver real time intelligence and information to frontline troops using an information network . Information about this new system is understandably not exactly full of details. The system concept, however, is pretty much in keeping with the long story of the rise and rise of better combat information systems.
Multiple levels of Squad X, notably a mobile system, are building a mix of ad hoc network and A.I. in to a credible model for combat purposes. That cannot possibly have been easy. Modern military information comes in Big Data sizes at command levels, and managing it has been a 20 year epic of note since the Second Gulf War.
To get to the point where data can be workable at small unit level is no minor achievement. However – This tech also opens a few dozen cans of worms, theoretically and operationally.
The need for reliable intelligence
Before anyone gets too righteous in any direction - Remember that frontline troops need both fast and reliable info, and that conventional intelligence and information have always been pretty much an each way bet. Intelligence can be right or wrong, and people’s lives depend on it. It’s highly debatable whether Squad X, or even an actual enemy, could do worse than some of the lousy information fed to frontline troops over the last century or so.
There’s another, much less obvious issue here. Having intelligence is one thing. What you do with it, and whether you understand it, is another, often lethal, problem.
To give an example – Once upon a time in Vietnam, an Australian battalion decided that despite reasonable intelligence to the contrary, a North Vietnamese regiment wasn’t in their area. It was. The Australians were sending out roughly a company’s worth of patrols, right in to the middle of this supposedly non-existent NVA regiment.
That response to intelligence could have totally trashed the company doing the patrolling. It could also have severely compromised the battalion itself in the face of a very close, very large, enemy force. After a long, truly horrendous fight, the Australians got out of the mess with support from New Zealand artillery and US airstrikes.
Information management caused this quite avoidable, very dangerous, situation. The command decisions and response and the intelligence about on the ground realities were total opposites. Troops on the ground often wonder about command responses, and that’s one of the reasons. The arrival of real time information systems for combat troops is a major improvement. The decision-making process at the grunts on the ground at least has something real to work with, for a nice change.
Fight by wire? Probably just as well.
You can also see why “reliable intelligence” is so important. Information, even using up to date systems, can lag behind combat conditions. A “get the hell out of there, right now” response can be delayed for too long by back-and-forth communications. Long, useless firefights can be very costly.
Fight by wire does make sense, on so many levels. In “asymmetric warfare” aka guerrilla or terrorist warfare, situations can change so fast that rear commands can’t keep up with them.
Meanwhile, the infantry have been accumulating information gathering systems like Christmas for many years now. The ever-useful micro drones used by the grunts to scout combat zones are a good example. Other info comes from a variety of sources, arguably too many sources, at different times.
Talking about “too many sources”- There was a classic case of a British unit provided with air cover and support in Afghanistan, as recorded in “An Ordinary Soldier]” by Captain Doug Beattie MC of the 1st Battalion Royal Irish Regiment. His unit’s supports were taken away right in the middle of a long, tough battle with the Taliban. Higher command priorities had shifted based on the tides of information, and the combat guys simply had to wear the results. (In fairness, the removal of support was done to the total disgust of the combat unit’s command, as well as the troops on the ground, in this case. Beattie’s book is highly recommended as a good insight into the realities of that war, by the way.)
Point being - Combat forces and command so often don’t (and can’t) have the same current information. That is an ongoing, serious problem; it gets people killed, and that’s what’s so important about Squad X. One of the major issues with all this hardware and sources of raw data is integrating the information and getting command and combat forces on the same page, literally. That is THE serious problem, even with modern communications systems, and Squad X may finally put a stop to it.
So…What’s wrong with the Squad X idea? Not much, but…
Now, the grunts can get their own information, in various forms, at battalion level. They have at least a chance of self-managing the new circumstances for themselves.
But - Feel free to groan at this point – The inevitable primary risk with Squad X is information security. If it’s hackable, the information can be compromised. That’s not fantastically good news for the people who will be depending on it. Jamming is also an obvious possible issue for information gathering. Fake sources patched from anywhere could be another difficult issue.
Another, even vaguer, issue is that “verifiable information”, the sort you might have to risk your life with, is also going to be relying on the efficiency of Squad X. That’s a much tougher ask for this general concept.
To explain:
• Hard information is what you know to be correct and current.
• Any other type of actionable information is open to interpretation and cannot be considered reliable until verified.
• “Electronic camouflage” is also not exactly unknown, and it’s critical information required ASAP. Verifying that sort of info is difficult at best.
It’s asking too much of A.I. to expect it to just wind up an algorithm or so and turn vague information into something useful. Basic deception tactics could be quite enough to derail it. Dummy missiles, for example, or even just lighting a few more fires at night, as done thousands of years ago, could do the trick. That’s “information”, but it’s useless, and consumes time and space.
Let’s not overdo it here – Somebody has to separate the fact from the fiction, and that’s what grunts do on a minute by minute basis. The problem is that a lot of misleading information can be fed into any system. Exactly how much of this unfettered garbage can be handled by Squad X in real time, in a real combat environment?
I think Squad X is a baseline necessity for future combat operations. It has the ability to adapt and evolve. It seems to be streamlining pretty well, too. The only real major issue I see is that there’s a systemic risk which has been built in to armies since the Pharaohs.
If information management is done at the expense of common sense, like, “The A.I. said so”, and it becomes the decision maker, the excuse and the justification, the SNAFUs are inevitable. History is full of military apologists who used “by the book” to excuse hideous failures.
This may be one of the first cases in history where grunts need formal training in applied skepticism as part of their standard combat situation assessments. It might be worth checking out that option, just to be on the safe side.
This opinion article was written by an independent writer. The opinions and views expressed herein are those of the author and are not necessarily intended to reflect those of
More about Darpa, Sex squad, Combat information systems, Combat intelligence, verifiable information
More news from Show all 6
Latest News
Top News