Does AI in warfare help the offense or the defense?

by on March 23, 2017 at 10:24 am in Web/Tech | Permalink

…I interviewed Eric Schmidt of Google fame, who has been leading a civilian panel of technologists looking at how the Pentagon can better innovate. He said something I hadn’t heard before, which is that artificial intelligence helps the defense better than the offense. This is because AI always learns, and so constantly monitors patterns of incoming threats. This made me think that the next big war will be more like World War I (when the defense dominated) than World War II (when the offense did).

Here is the link, by Thomas E. Ricks, via Blake Baiers.

1 anon March 23, 2017 at 10:43 am

OK, I’ll go there.

Is that the future of war? Or is getting a clown and appeaser elected on the other side the future of war? Maybe Putin really thought outside the box, and while we were worrying about drones in squad situations, he was clearing the board.

White House Ethics Lawyer for George W. Bush, Richard W. Painter: “FBI uncovering evidence of treason. There is no other word for it.”

http://www.mcclatchydc.com/news/politics-government/white-house/article139695453.html

2 Pensans March 23, 2017 at 10:48 am

The attested attack of Putin was on the credibility of the system, the release of DNC emails. Apparently, it worked. Truth has that effect.

3 anon March 23, 2017 at 10:53 am

There is an elephant standing in the room, and it certainly is not whether DNC members liked Bernie Sanders on a personal basis.

““This may be one of the most highly impactful information operations in the history of intelligence,” said one former U.S. intelligence official, who spoke on condition of anonymity because of the sensitivity of the matter.”

That Sirs, is the future of conflict in an information age.

4 MOFO March 23, 2017 at 11:04 am

I havent been keeping up with this latest pants shitting, are we still talking about the DNC emails? Are Dems still claiming, in essence, that Hillary’s campaign was so weak and inept that she could not beat Donald Trump (!) after the leaking of a slightly bad email by someone that wasnt even part of her campaign?

There have been so many vacuous claims by the losers in this last election that i cant even keep them straight anymore.

5 anon March 23, 2017 at 11:09 am

Maybe you should keep up. Look for the word “treason” above.

I have been slow to comment on the Russia stuff because it does seem so cloudy and without obvious logic, but never ever in the history of Hillary leaks did anyone use the word “treason.”

Something flipped, and maybe it’s time for you to do some homework.

6 MOFO March 23, 2017 at 11:29 am

From the article you cited above:

“Operatives for Russia appear to have strategically timed the computer commands, known as “bots,” to blitz social media with links to the pro-Trump stories at times when the billionaire businessman was on the defensive in his race against Democrat Hillary Clinton, these sources said.

The bots’ end products were largely millions of Twitter and Facebook posts carrying links to stories on conservative internet sites such as Breitbart News and InfoWars, as well as on the Kremlin-backed RT News and Sputnik News, the sources said”

Yes, truly a scandal for all times.

7 anon March 23, 2017 at 11:31 am

The “bots” are of course the parallel to “AIs” above, but that’s not where the treason comes in. That happens here:

“Investigators examining the bot attacks are exploring whether the far-right news operations took any actions to assist Russia’s operatives.”

You’d think “kids today” would know that working with foreign powers to influence an election was treason, but maybe our public schools fail.

8 Cliff March 23, 2017 at 12:26 pm

“working with foreign powers to influence an election was treason”

So you’re saying he treason was committed by the news outlets?

And… what? How could that possibly be treason? Don’t tons of foreigners donate to Super PACs?

9 MOFO March 23, 2017 at 12:42 pm

The treason here is that Hillary could never win in the face of anything other than a totally obsequious and fawning press. Having automated accounts say mean things about her on facebook and twitter all but guaranteed that her weak and inept campaign would implode.

10 anon March 23, 2017 at 1:00 pm

“So you’re saying he treason was committed by the news outlets?”

Where was Bannon working?

11 Anonymous March 23, 2017 at 3:03 pm

““Investigators examining the bot attacks are exploring whether the far-right news operations took any actions to assist Russia’s operatives.””

And they won’t find anything, you’ll forget it next week and make a big stink about the next “investigation.” Fake News.

12 anon March 23, 2017 at 3:38 pm

I can’t remember having to reverse, but something sad is going on:

https://morningconsult.com/2017/03/23/trump-scandals-make-voters-like/

If it turns out the campaign did collude with a foreign power, how many of you will just decide you like that?

13 Mark Bahner March 23, 2017 at 5:22 pm

“You’d think ‘kids today’ would know that working with foreign powers to influence an election was treason, but maybe our public schools fail.”

OK, I’m not a kid, but I can read the Constitution: “Treason against the United States, shall consist only in levying War against them, or in adhering to their Enemies, giving them Aid and Comfort.”

How does “working with foreign powers to influence an election” fit the Constitution’s definition of treason?

14 Thomas March 23, 2017 at 8:11 pm

It’s not treason but anon ‘has been very slow on Russia’, and let me remond you that he came to MR as a moderate. Lots of moderates read thinkprogress daily and accuse Presidents of treason.

15 TallDave March 26, 2017 at 3:54 pm

The returns from better military planning are probably minimal at this point, and remember the military also learns, institutionally speaking, from AI. The issue with AI is not that, it’s the possibility of manufacturing mass numbers of of autonomous weapons. That should really worry people. Nukes at least would leave some survivors, the hoary sci-fi trope of a planet left utterly lifeless by such machines is unlikely but not so totally remote it doesn’t merit a lot of preventive worry.

Anyways, fortunately there likely won’t be any such machine armies or a Next Big War, as there’s not much left to fight about in the age of liberal democracy, just periodic death spasms of states everyone already knows are failed. The Islamic Caliphate currently collapsing around Raqqa might get a couple more go-rounds, but that seems to be self-limiting because it’s so fundamentally stupid. Really all that’s left is to clean up the corruption, get rich, and help humanity work toward larger goals like terraforming, curing aging, and colonizing the planets and stars. All of these might be achievable in the next 100 years if world living standards approach the equivalent of $100K per capita today.

16 TallDave March 26, 2017 at 3:55 pm

Sorry, that wasn’t supposed to go there.

17 Captain Obvious March 23, 2017 at 1:03 pm

If treason in deed occured, then this is the most spectacular failure of US intelligence since ever. Beaten by a country with GDP of Italy. This does not pass the logic test.

18 Tom Vansant March 23, 2017 at 1:52 pm

The beauty of the electronic age is that it takes minimal resources.

I would agree with the origional post. This was a brilliant attack by an advasary. So good in fact that the dupes who waged the attacks are not defending their own actions. I don’t know if it was Putin or Maniford but it was certainly treason.

19 Mark Bahner March 23, 2017 at 6:17 pm

“I don’t know if it was Putin or Maniford but it was certainly treason.”

From the Constitution: “Treason against the United States, shall consist only in levying War against them, or in adhering to their Enemies, giving them Aid and Comfort.”

Where is the treason?

20 Rich Berger March 23, 2017 at 11:29 am

In the race for most tedious and tendentious commenter, anon and prior_test are neck and neck.

21 anon March 23, 2017 at 11:33 am

I’ll say it again. This is directly on topic. It is the cyber war, bots and all, that just happened.

22 Cliff March 23, 2017 at 12:29 pm

There’s no AI and no warfare so actually it is completely off topic

23 anon March 23, 2017 at 12:36 pm

It is always best to attack an undefended flank. Imagine doing that so well that your enemy doesn’t understand the scope of your effort until months after your victory.

This isn’t done, the FBI may give Trump a clean call with caution, as they did Hillary.

But even that, a caution about running a campaign without foreign influence, would be pretty damning.

24 Jeff R March 23, 2017 at 1:13 pm

But here’s the courageous anon to rescue us all from our own ignorance. A shining beacon, bringing to light the depths of our nation’s manipulation and victimization by that awful Vladimir Putin. Why, they ought to give you the Congressional Medal of Freedom.

25 MOFO March 23, 2017 at 12:44 pm

Ok then, ill say it again, “war” in this context is “The bots’ end products were largely millions of Twitter and Facebook posts”

The horror, the horror.

26 anon March 23, 2017 at 12:54 pm

Are you sure “bots” will be the only thing in Coney’s report? No flow of funds? No coordination of principals?

Don’t make your bed too soon.

27 MOFO March 23, 2017 at 12:58 pm

The minute i see something actually noteworthy, ill happily change my tune. So long is it continues to be mean Facebook posts and emails from the DNC, ill continue to mock you desperately grasping for anything to excuse your loser of a candidate’s pathetic performance.

28 anon March 23, 2017 at 1:02 pm

I quoted White House Ethics Lawyer for George W. Bush, Richard W. Painter above.

I am happier to have him in my corner, rather than some MOFO.

29 MOFO March 23, 2017 at 1:09 pm

So in the post Trump world, George W Bush’ ethics lawyer (!) is now a hero? I guess whatever it takes to make super mean tweets into treason, amirite?

30 prior_test2 March 23, 2017 at 12:10 pm

And to think that our gracious hosts ensure you cannot even read everything I post here.

31 msgkings March 23, 2017 at 1:18 pm

If only they’d finish the job…

32 Anonymous March 23, 2017 at 3:05 pm

+1

33 Jeff R March 23, 2017 at 3:51 pm

Lulz.

34 anon March 23, 2017 at 11:29 am

I’m sorry I have to explain something so simple, but maybe some of you need a parallel.

You have called Hillary a “socialist.” Imagine that the Communist Chinese saw her as a socialist, for reals. Imagine that they started a cyber war, an information war to support her. Imagine that they hacked the Republican Party. Imagine that Hillary “the socialist” had her campaign coordinate with them. And then she won.

How would you feel about that? Would you be as comfortable with Communists hacking an election as you (apparently) are with “strong men” doing it?

35 MOFO March 23, 2017 at 12:39 pm

And by “hacking an election” he of course means coordinating some mean tweets and maybe possibly having something to do with the release of emails from a tangentially related third party.

36 msgkings March 23, 2017 at 1:19 pm

His hysteria and narcissism does mask a good point, MOFO. If this happened to the Reps you’d be frothing at the mouth about it.

37 MOFO March 23, 2017 at 1:55 pm

Really? You know me well enough to know that?

Heres the actual truth: I used to do just that. After a couple of decades of carrying water for whatever clown happened to be nominated for election that year, I got tired of the mental gymnastics it takes to vociferously argue a point, then just as vociferously argue the opposite point as required by the circumstances of the day. I know its hard to believe, but there are people who arent mindless partisans. “the other side does it” isnt always a dead-bang backup position.

38 msgkings March 23, 2017 at 1:58 pm

I’m glad you aren’t a mindless partisan, but that makes you a pretty rare bird on this site. I was just playing the odds.

39 Anonymous March 23, 2017 at 3:07 pm

There’s no hypothetical about it, communist propaganda/subversion, directed from the Soviet Union, was a reality for decades.

40 anon March 23, 2017 at 3:27 pm

Right. The people who *aren’t* mindless partisans want to bury uncomfortable news.

http://www.thedailybeast.com/cheats/2017/03/23/another-intel-dem-confirm-trump-russia-evidence-more-than-circumstantial.html

41 CorvusB March 23, 2017 at 6:32 pm

@Mark Bahner: Treason?
in adhering to their Enemies, giving them Aid and Comfort.

right there.

Hillary doesn’t matter a whit in this. She was an awful candidate, in spite of the fact that her husband was a very good president. She was no Bill Clinton.

Regardless of all that – the extent of the REAL fake news programs against the Dem side were not unlike the Russian news campaign against Ukraine in the early days of that war. For those of you who are dismissing the danger here, you are not up to date on current Russian history. Go back over the past 15 years and look at events and techniques surrounding Russian political events. Like the Yanukovych election. Look at the Georgia war – what was attempted, and what was accomplished. Now look at the Ukraine war – and look at what has been accomplished. All with a minimal exertion on the part of the Kremlin. Cagey. Clever. Looks very much to me like someone wants to be this generation’s Stalin. (And many Russians welcome that. I understand their reasoning – Russia WAS a superpower, and they want that back. I just think they are on a very wrong track.)

For those of you who would dismiss the danger of this external campaign, you are guilty of working AGAINST liberty and democracy – and the values embodied in the creation of these United States. I have watched the pendulum swing for several decades, and we have now swung into the most dangerous grounds that I have seen in my lifetime. I do not think the 60’s even made it this far. This is not “mean tweets”. This is all-out nationalist based trolling and disinformation. And it has been exceptionally effective to date.

42 Mark Bahner March 23, 2017 at 6:45 pm

“@Mark Bahner: Treason? in adhering to their Enemies, giving them Aid and Comfort.”

Russia is our enemy? So aiding their government is treason?

43 Mark Bahner March 23, 2017 at 7:02 pm

“Russia is our enemy? So aiding their government is treason?”

The reason I ask about this is that the US Chairman of the Joint Chiefs of Staff, Joseph Dunford, met with the Russian Chief of General Staff Valery Gerasimov earlier this month…and I don’t remember anyone claiming this was “treason.” Wouldn’t working with the head of an enemy’s military be treason?

Also, NASA is paying the Russian government about half a billion dollars to fly U.S. astronauts to the Space Station. Wouldn’t paying an enemy government half a billion dollars be treason?

44 Mark Bahner March 23, 2017 at 6:14 pm

“Imagine that they started a cyber war, an information war to support her.”

What is an “information war”? Is it like the Monty Python Lethal Joke sketch? People see the information and die?

“Imagine that they hacked the Republican Party.”

I would not be surprised if the Chinese government are already “hacking” both parties. (And I definitely would not be surprised if the Chinese hacked Hillary Clinton’s email server when she was Secretary of State. And I’m sure, if Donald Trump is communicating by some sort of “private secure” means, it’s not private or secure.)

“How would you feel about that? Would you be as comfortable with Communists hacking an election as you (apparently) are with ‘strong men’ doing it?”

I don’t understand what “hacking an election” means. Do you mean like altering the results of voting machines?

45 Jeff R March 23, 2017 at 11:32 am

Maybe get a hobby, guy.

46 msgkings March 23, 2017 at 1:20 pm

This, of course, is his hobby. Same as the rest of us.

47 Jeff R March 23, 2017 at 3:52 pm

A different one, then.

48 So Much For Subtlety March 23, 2017 at 1:49 pm

Thanks to good old King George III treason in the US constitution and law has a specific meaning.

Thus it did not apply when Jane Fonda sat on an anti-aircraft gun and praised the people torturing and killing US servicemen. It did not apply when Fonda and Tom Hayden took Hanoi’s (and hence Moscow’s) money to oppose a morally just war.

It does not apply here. The idea that some people “associated” with Trump talked to some other people “associated” with Moscow is hardly proof the Russians hacked the election. In fact, I can’t see what is wrong with talking to John Podesta at all.

49 anon March 23, 2017 at 2:58 pm

I have one more .. wait for it .. Yes this was an unexpected form of asymmetric warfare, and everyone who defended it became a Fifth Columnist – in the original meaning of the term.

https://en.m.wikipedia.org/wiki/Fifth_column

50 Doug March 23, 2017 at 3:57 pm

How exactly is hacking into a political party’s servers and distributing their e-mails equivalent to warfare against a country? Is the DNC an arm of the United States government? No, of course it isn’t–although it seems like some democrats think it should be. Does the United States government have a specific interest in the outcome of the 2016 presidential election? No, the government does not get to vote. Were the e-mails forgeries? I haven’t seen any evidence of that. Is there any evidence that Donald Trump is some sort of sleeper agent, with secret loyalties to our enemies? Again, the Democrats seem to think he is, but I have not seen any evidence that he is giving away our launch codes, or engaging in any actual treasonous acts.

So what exactly are you complaining about? Some truthful information about your candidate and your party got out, and as a result, they lost the election. Somebody call the waaambulance.

I would say that calling everyone who disagrees with you a traitor and a fifth columnist is the new form of McCarthyism, but in this case it is much more pathetic than scary.

51 anon March 23, 2017 at 4:08 pm

I am not the Washington and GOP insider and expert who used the word “traitor.” I did not launch an FBI investigation into collusion with an enemy power. I am not the Senator who said there is more than circumstantial evidence.

Start by recognizing those as external facts.

Once you do, and you recognize the Russian win, then yeah, I think my coinage, that this was a successful cyber-war makes sense.

52 anon March 23, 2017 at 4:13 pm

And beware, in case it isn’t obvious, a green light for Trump and the Russians in 2016 is a green light for Democrats and the Chinese in 2020.

That is what happens when you forfeit national sovereignty.

53 MOFO March 23, 2017 at 4:21 pm

So the facts that we should ‘recognize” are the opinions of two politicians and the fact that there is an investigation? And to be clear here, we are still talking about facebook posts, tweets and leaked emails from a third party unrelated to Clinton, right?

54 MOFO March 23, 2017 at 4:24 pm

“And beware, in case it isn’t obvious, a green light for Trump and the Russians in 2016 is a green light for Democrats and the Chinese in 2020.”

So are you saying that we can expect to be bombarded with Facebook posts and tweets supporting the Democratic nominee for president in 2020? What will that be like, i wonder? What kind of horrible vision of the future are you offering?

55 anon March 23, 2017 at 4:30 pm

It doesn’t serve any purpose to play dumb. Trump invited Russian intelligence to hack his opponent.

http://www.politico.com/story/2016/07/trump-putin-no-relationship-226282

If you understand that now, and give that a green light now, you (certainly not I) are giving Chinese intelligence a green light to hack Republicans.

Me? I’d like to go back to US sovereignty, and cyber protection against all foreign intervention.

56 MOFO March 23, 2017 at 4:33 pm

“US sovereignty”. Wow, when you put it that way it sounds way more serious than a bunch of facebook posts and tweets, doesnt it?

57 Mark Bahner March 23, 2017 at 5:42 pm

“Once you do, and you recognize the Russian win, then yeah, I think my coinage, that this was a successful cyber-war makes sense.”

Don’t we have to actually be at war for a “successful cyber-war” to occur?

58 Doug March 23, 2017 at 5:59 pm

I suspect the Chinese are already hacking or attempting to hack into the e-mails of both the republican and democrat parties, and have been virtually since the invention of e-mail. Is it a problem? Yes. Is it warfare? Not even close.

59 Mark Bahner March 23, 2017 at 5:49 pm

“I have one more .. wait for it .. Yes this was an unexpected form of asymmetric warfare, and everyone who defended it became a Fifth Columnist – in the original meaning of the term.”

In the original meaning of the term, Fifth Columnist was applied relative to the Spanish Civil War. Do you think the 2016 presidential election was part of a U.S. civil war?

60 Mark Bahner March 23, 2017 at 5:09 pm

“Is that the future of war? Or is getting a clown and appeaser elected on the other side the future of war?”

We’re at war with Russia? When did Congress declare war on Russia?

61 Tim C March 23, 2017 at 10:44 am

The World War One analogy seems like a poor interpretation of the phrasing. AI could help the “defense” (I disagree with the statement, but whatever) in the sense that the side that is attacked could benefit more from it, learning and responding to the attack, but *how* each nation “attacks” and “defends” wont look at all like World War One, which was product of offensive tactics being slow and costly. Nothing here suggests war wont look like quick, massive destruction from missile-type weaponry, just that whoever attacks first wont have as strong an advantage.

62 Tzu March 23, 2017 at 12:50 pm

[]

[ ” ….artificial intelligence helps the defense better than the offense..” ]

Nonsense. “Better” intelligence helps whichever side has the better intelligence.

That key ‘intelligence’ is the size/capability/disposition/intentions of the opposing forces, regardless of whether they are currently in a defensive or offensive posture.

The status of defensive or offensive can change quite quickly– and can be a complex mixture of both at dynamic tactical levels.

Basics of warfare have been the same for thousands of years.

Typically, a prepared defensive force is superior to an attacking/offensive force of the same size/capability. Military rule of thumb is that an attacking force should be at least 3 times the size of the defenders in order to offset their inherent advantage.
Many factors affect the balance, including intelligence available to the combatants.

Game Theory states that a sharply aggressive attacker enjoys an added advantage over a defender… if both lack good intelligence about their opponent.

That WWI reference is irrelevant to intelligence; defense advantage was primarily due to improved weapons (machine-guns/artillery).

63 Ray Lopez March 23, 2017 at 11:18 pm

Finally, a smart comment from a student of war history. I’ll add that besides offense being slow and costly (the artillery has to be accurate when working with infantry to within well under 100 meters, and there was no way to do this before wireless very accurately), the defense was not appreciated in WWI, hence the battles of Verdun, inter alia, to ‘bleed dry’ the enemy. The thinking was offense is good since it will result in attrition. Nasty thinking that did not heed defense at all.

Bonus trivia: the more mobile WWII was in some ways more bloody than WWI, as more people died over a wider area. Not sure about per capita soldier deaths however, which typically hover around 10% unless you’re talking about the Soviet military (or Turkish in WWI), where it’s much higher.

64 Alistair March 24, 2017 at 6:33 am

Ray,

From another student of war (well, operations research). The critical metric for comparison here is deaths per solider-day. WWII killed more people, but only because it was bigger in time and scale, not deadlier on an intensity basis (at least not for Western Troops. German, Soviet, and Japanese experience may differ).

I would say the defence-offence balance is also oft misunderstood. The 3:1 rule one of those heuristics which has exceeded it’s utility; it’s actually only true for a relatively small subset of encounters. As Tzu says, other considerations usually dominate. Even in WWI people focus on the stasis of the Western Front 1915-17, and ignore the successful offensives and mobility in the East, the middle east, and the 1918 campaigns….

65 dearieme March 23, 2017 at 10:49 am

The crucial difference between the two WWs was in command-and-control. It’s got better since WWII unless electronic jamming takes it all backwards.

Still the suggestion that we’ll get a reprise of WWI must hearten the proponents of the F-35: it might succeed there.

66 So Much For Subtlety March 23, 2017 at 2:01 pm

Except command and control is so good it has got worse again. Ever since the Vietnam war, any commander can expect to have his commander on his back – a junior officer fighting on the ground might have three layers of command in the air above him, all second guessing what he is doing. If he is really unlucky he might have the President on the line demanding to know what he is doing.

There is too much c’n’c. Most wars would be better with less.

67 Bob March 23, 2017 at 10:51 am

That’s a good extrapolation on how it works for fraud detection but there’s one key difference: When it comes to fraud, the number of datapoints is crazy, and companies are OK with having high fraud for a while and get better at countering it as the training sets get bigger. In war, people come at you with their best punch first, hoping to just win.

If the US was going to have to enter a defensive war of any sort, I suspect that the first hour would involve an attacker doing whatever is possible to stop the US from ever fighting back: No time for AI, or humans for that matter, to learn anything.

68 Melmoth March 23, 2017 at 11:03 am

Yes, AI ‘learns’ on thousands of data points or more, i don’t see how it can effectively learn from one event.

69 mkt42 March 23, 2017 at 4:53 pm

Correct. And AI has another limitation due to its need for training data: it’s inherently backward looking. Deep Blue could play a mean game of chess, but suppose after it beat Kasparov it would’ve had to play poker against its next opponent?

During WW I France learned how to build and defend static fortifications to the state of the art. But in 1940 the Germans played poker instead of chess, using armor, mobility, modernized command & control, and blitzkrieg tactics.

Clearly humans can have trouble anticipating their next opponent just as AI does. But humans also have greater imaginative capabilities (computers can write novels now, but they can’t write good ones). The Japanese Navy in WW II was stuck in a WW I mindset and thought its carefully husbanded fleet of battleships would fight the decisive battle, not realizing that the game was now poker and aircraft carriers were the key weapon (along with submarines for strategic warfare). Yet that same hidebound Navy was imaginative enough to create and use an unprecedentedly large and high quality aircraft carrier fleet, which led it to a string of victories in the six months after Pearl Harbor.

How can AI be trained for a mode of warfare that hasn’t yet been fought?

70 Axa March 23, 2017 at 10:52 am

That pattern looking reference sounds like counterintelligence to catch foreign agents.

71 Hadur March 23, 2017 at 10:57 am

Let’s see what those AIs are capable of after being hit with an EMP

72 JWatts March 23, 2017 at 12:31 pm

It would seem pretty trivial to harden any expensive AI against EMP. A grounded metal cage will work just fine. Your standard heavy duty electric drill is EMP proof.

73 Thanatos Savehn March 23, 2017 at 11:10 am

The foundations of statistical inference deployment in warfare: ordnance quality control, reinforcing fighters/bombers and predicting enemy next move from bombing patterns (among others) suggests that machine learning will primarily accelerate the already rapid increase in lethality of U.S. forces.

74 collin March 23, 2017 at 12:29 pm

My guess is defensive positions improve because the destruction of modern nations is so expensive and costly today. Isn’t it true the rich and more powerful nation since WW2 generally lose modern wars because with AI & technology they really hate to sending actual troops to the location. Look even Iraq 2 which really could be called a US victory* had the reality of an empty victory and it cost a bunch money and blood for nothing. (Trump winning the election with the Take Their Oil Populism supports this theory!)

*The objectives of Iraq War was to depose Saddam, stop the WMD programs, create a democratic govenrment and a peaceful thriving economy. We achieved the first three objectives so I recommend we call it a victory instead of a loss. I also believe calling it a victory makes the war feel a lot more Empty than a loss where we could play Rambo ten years later.

75 JWatts March 23, 2017 at 12:29 pm

I think it’s impossible to predict whether AI will trend towards enhanced offense or defense at this point.

However, the big change on the horizon is high powered Lasers. We rapidly approaching the thresholds whereby Lasers have enough power to destroy aircraft of all types. This will fundamentally change warfare as we know it. High powered lasers (with rapid charging), coupled with cheap and omnipresent sensors will essentially change warfare into the “It flies, it dies” category. And since the very same Lasers will probably be equally capable against mortars and artillery, almost all air borne weaponry would be significantly degraded.

This would lead us to a very defensively strong military balance, with only heavily armored ground vehicles capable of offensive actions in open terrain.

76 MOFO March 23, 2017 at 12:48 pm

Maybe, but lasers arent impossible to harden against either. Still, the “it flies it dies” concept might end up being true due to cheap(er) high speed anti aircraft missiles and so on as well as lasers.

I gotta think it would take a really really powerful laser to cut through the skin of an artillery shell inside the engagement window offered.

77 Turkey Vulture March 23, 2017 at 1:06 pm

Hammer’s Slammers.

78 Patrick Laske March 23, 2017 at 1:52 pm

This is closer to the topic at hand, in that AI will help lasers defenses work more efficiently to deal with incoming missile attacks, but the mark is completely off. It would be incorrect to say this enhances defense though, it actually increases force projection in making carriers and heavy ships able to operate closer to shore and project power. The limits of offensive capability is your opponent’s ability to deny you access through cheaper and more powerful weapons as you get closer to their home (which has the fancy name Anti-Access Area Denial or A2/AD).

The problem with using laser weapons is that there’s a limit of about 10-20 miles to hit a target if you’re on the surface. Problem of living on a sphere. Even old Anti-Ship missiles used by Chinese planes have a range of 75 miles, with modern plane launched ones with a range over 300 miles.

Lasers also don’t work like they do in movies. They require a huge amount of power and a realistic time on target to melt it. Like 5 seconds for a few hundred KW laser. Right now they’re testing lasers in the low hundred thousand KW range, mostly because that’s how much extra power there is in the current set of ships. Newer Destroyers and Carriers that will come down in the future will have powerplants capable of projecting more power. Planes tend to move around and made of sturdy material making them annoying to melt, so you’re basically looking at needing power in the 1-10 MW range to realistically melt them. Ship lasers that big require power plants in the 20 to 100 MW of power and displace about 1,500 to 15,000 tons of coolant. You could build a boat today that could do that, or static defenses, but it’s unrealistic for a plane and it’s unrealistic against supersonic jet planes.

MOFO is correct that you can harden against lasers, (or make them faster or make them more maneuverable or harder to detect) but they would be effective against preexisting stockpiles of anti-ship missiles – which cost a bunch of money to acquire and took decades to build up.

Another way of thinking of Offensive Warfare vs Defensive Warfare is Expensive Wars vs Cheap Wars. AI is about replacing expensive human labor (thinking) with cheap faster computers. Economics and scalability play a big role in offense vs defensive dominant technologies. When war is cheap, you get more war. Stronger defenses (a castle) make war more expensive (going to have to siege for like 3 years to break this castle). As such, I’m going to overall lean on AI helping offense more than defense. In 50 years thanks to robots, I expect the difference between the 150th strongest military and the 25th strongest to be quite small, though you might see some asymmetries in the high tiers still.

79 Lord Action March 23, 2017 at 2:18 pm

You don’t need to power lasers continuously. Even assuming you’re using diode lasers, the power comes from some capacitor bank that’s charged at a low average power. And the really big lasers are, so far, chemical and so require essentially no electricity.

80 JWatts March 23, 2017 at 5:25 pm

“The problem with using laser weapons is that there’s a limit of about 10-20 miles to hit a target if you’re on the surface. Problem of living on a sphere. Even old Anti-Ship missiles used by Chinese planes have a range of 75 miles, with modern plane launched ones with a range over 300 miles.”

That’s what makes laser weapons primarily a defensive weapon. They’ll stop incoming missiles/aircraft but they won’t project power very far. And those anti-ship missiles are now worthless, unless you can saturate the defenses.

“Newer Destroyers and Carriers that will come down in the future will have powerplants capable of projecting more power.”

Yes, the new Destroyer USS Zumwalt has around 80 MW of electrical power generation, though that is shared with the propulsion system.

“Lasers also don’t work like they do in movies. They require a huge amount of power and a realistic time on target to melt it. Like 5 seconds for a few hundred KW laser. ”

Yes, but it’s close to a linear scaling. So, if you need 5 seconds for a 100 KW laser, then you only need 0.5 seconds for a 1 MW laser and only 50 mSecs for a 10 MW laser.

81 Alistair March 24, 2017 at 6:40 am

+1 for describing the LDEW/CRAM problem succinctly.

82 Yancey Ward March 23, 2017 at 12:33 pm

WWII was the result of the learning from WWI.

83 Turkey Vulture March 23, 2017 at 1:05 pm

“AI always learns, and so constantly monitors patterns of incoming threats.”

Perhaps in a bygone era this meant it was “defense-favoring.” But nowadays, and certainly ever moreso in the future as AI capabilities grow, the offense will want to “constantly monitor patterns of incoming threats,” i.e. interceptors of various sorts.

So the defense gets better at targeting the offensive unit, and the offensive unit gets better at avoiding what the defense is targeting it with (or better at counter-targeting those defensive units with its own defenses – turtles all the way down).

The more interesting issue to me is whether it favors the masses or the powerful elite, and/or the resource-poor but committed insurgents against the rich but not-as-committed foreign power. So I wonder: how much better will the pricey version of whatever future weaponry we are imagining be than the cheap version, and can the cheap version successfully inflict casualties?

84 rayward March 23, 2017 at 1:40 pm

“This is because AI always learns, and so constantly monitors patterns of incoming threats.” A good offense requires the unexpected; a good defense plans for the unexpected but anticipates the expected (and, hence, requires more resources, the most efficient allocation devoting the most to the expected).

85 albatross March 23, 2017 at 2:19 pm

If our societies are vulnerable to a lot of disarming first strikes that human minds aren’t good st finding, you could imagine AI being extremely destabilizing. Like flame throwers at two paces.

86 Mark Thorson March 23, 2017 at 3:40 pm

If AI makes it possible to fight a war with drones and robots alone, that will considerably lower the bar to engaging in war. When you don’t have people in the combat theater, you don’t have bodies coming home and grieving families (not here, anyway). Want to remove Mugabe from power? Send in the robots. Want to stop a genocide? Send in the robots. Want to replace a homophobic, sexist government? Send in the robots.

87 albatross March 24, 2017 at 12:49 pm

This is more-or-less how we’re using drones, right?

88 roystgnr March 23, 2017 at 3:41 pm

AI always learns when provided with new data. If an enemy’s defensive capabilities and planning (which must be continuously updated and practiced) are easier data to acquire than an enemy’s offensive capabilities and planning (which can be ramped up right before initiating a war and so are vulnerable to intelligence operations for a shorter time), then the offense gains a greater advantage from AI.

89 Chip March 23, 2017 at 4:56 pm

We’re reaching a point where AI will do almost everything quicker, if not better, than people and governments may decide that to keep up with rivals they will have to remove all restraints on their native AI – and hope it has enough built-in sympathy to fight for their interests.

So a China-built AI with Chinese characteristics such as an affinity for authoritarianism will face off against an American AI that’s geared toward decentralisation and individual liberty.

We people will become mere spectators of what will be a very brief conflict.

90 Rafael R March 23, 2017 at 5:22 pm

The next war between major powers will be at sea (China, US pacific). So it will be completely different from European ground warfare that characterized WW1 and WW2. Main difference between WW1 and WW2 was that in WW1, the Western front was “static” because each inch of ground was precious so we had about 200 divisions on each side fighting over tiny pieces of land, in WW2, the main front was the Eastern front, where you had thousands of kilometers of frontage and millions of square kilometers under dispute, so warfare was more “mobile”. While in 1940 and 1944-45, in the Western front, in both cases, one side had overwhelming superiority over the other so it was relatively quick. It’s not comparable to 1914-18, when the W. front had relative numerical parity.

91 JWatts March 23, 2017 at 5:35 pm

You are confusing cause and effect. WW2 started with a Western Front with vast well prepared defensive positions on the Allied side, but German blitzkrieg tactics walked right through the Belgium defenses. They quickly maneuvered behind the Maginot line and entrapped most of the French army (and the British got away by the skin of their teeth). The Germans were just as effective versus the Russians. It was the mud of fall and the cold of winter that brought the Wehrmacht to a halt after driving the Russian Army back 1,000 miles.

92 bellisaurius March 23, 2017 at 6:21 pm

Why would AI be any different at handling things then a thinking person in the long run? They’d have access to the same sort of decision matrices.

93 Troll me March 25, 2017 at 2:59 pm

The logic assumes that no one is transmitting all of this AI-learned defense directly to the enemy.

The risk that one rival may calculate that they can successfully overcome defenses at an acceptable cost may be much higher under such conditions.

Uncertainty in this regard could legitimize Trump’s preference to load up on nukes. Which I think is a highly suboptimal use of resources. But there’s something to that.

94 AlexC March 26, 2017 at 2:04 am

Few people commenting on this story have any idea what they are talking about. The only aspect of War that AI will have an immediate effect on is the implementation, effectiveness, and lethality of EW, ECM, ECCM, EPM, IA, PSYOPS, AESA’s, and everything else that fits this theme.

95 TallDave March 26, 2017 at 3:56 pm

The returns from better military planning by AI are probably minimal at this point, and remember the military also learns, institutionally speaking, from AI. The issue with AI is not that, it’s the possibility of manufacturing mass numbers of of autonomous weapons. That should really worry people. Nukes at least would leave some survivors, the hoary sci-fi trope of a planet left utterly lifeless by such machines is unlikely but not so totally remote it doesn’t merit a lot of preventive worry.

Anyways, fortunately there likely won’t be any such machine armies or a Next Big War, as there’s not much left to fight about in the age of liberal democracy, just periodic death spasms of states everyone already knows are failed. The Islamic Caliphate currently collapsing around Raqqa might get a couple more go-rounds, but that seems to be self-limiting because it’s so fundamentally stupid. Really all that’s left is to clean up the corruption, get rich, and help humanity work toward larger goals like terraforming, curing aging, and colonizing the planets and stars. All of these might be achievable in the next 100 years if world living standards approach the equivalent of $100K per capita today.

Comments on this entry are closed.

Previous post:

Next post: