
Given the accuracy of the human targetters used hitherto, they wouldn't have noticed any decrease in efficacy. What was the hit rate again, 85 civilians killed for every militant neutralised?
The US Department of Defense has deployed machine learning algorithms to identify targets in over 85 air strikes on targets in Iraq and Syria this year. The Pentagon has done this sort of thing since at least 2017 when it launched Project Maven, which sought suppliers capable of developing object recognition software for …
Even if your intelligence is good, selectively killing individual people while protecting other people in the vicinity is very hard to do when the chosen weapon is a high-explosive bomb dropped from an aircraft. The only way you can really do is to send in soldiers on the ground, but that brings a whole different set of problems.
Targeting system have the accuracy to allow non explosive weapons, a few hundred kilos of cement filled bomb case or bladed hellfire missile can destroy a single vehicle and barely scratch the next car along. It'd be nice to think they also try to have the ballistic path such that a guidance failure won't result in any photogenic screw ups.
Of course this has no benefit for any misidentified targets.
...the AI bombs Damascus, Maryland and Bagdad, Florida? [Yes, they do spell it like that there.]
You can understand why they are trying to up the success rate. Despite the promises, the Houthi Environmental Alliance are still taking down the West's consumer economy, one container ship at a time. Eventually they will take out another container full of yellow plastic ducks, which will float around the world and be used as evidence of our pressing need to switch to bespoke wooden keyboards from Etsy.
Perhaps the Metaverse went wrong and we are all trapped inside a Mel Brooks movie.
Killer robots will become a reality. Why? Because the U.S. DoD is too afraid China will built them first and thereby gain a military advantage. Ergo it's therefore a foregone conclusion they will be built and most likely are already being developed.
And it wouldn't surprise me in the slightest if the U.S. were to deliver them to Ukraine in the near future.
Missiles are too expensive and can only be used to attack high-value targets (ammunition and fuel storage, factories, command and communications centers). We only have a very limited supply and can't easily make more (and the government is unwilling to pay for huge stocks) and lead times are high.
We've seen that in Ukraine attacks on high-value targets haven't really changed the situation on the ground since we don't allow Ukraine to attack targets in Russia. The Russians still hold their ground and you need enormous amounts of "dumb" ammunition and manpower to break the deadlock.
Killer robots could be used to break the deadlock, if they work as advertised.
I am still not sure how a country can justify killing people for being criminals, where there has been no trial, by sending weapons of war into the air above a nation that they are not at war wiith. How is that legal or accepted. If Putin was doing it to kill pussy riot members in Europe, there would be no end to the shouting by Western leaders. Allowing AI to make the decisions does not make me feel any better. It would probably go after one of Googles Black founding fathers from the US who would certainly not approve of doing such things.
I don't think your argument works in this case when the Houti de facto government is firing missiles to attack shipping - that IS war.
The question is why is Israel demonstrating so much Brawn over Brain behaviour, and why does the US let that slide? From the decisions not to have a proper border guard by Gaza, to ignoring the intelligence that warned of an attack, to a response that leaves Gazans no way to avoid death, not even opening refugee camps, ultimately allowing the Houti, supplied by Iran, to act without reaction from Arab neighbors who really don't like them on the whole.
"The question is why is Israel demonstrating so much Brawn over Brain behaviour, and why does the US let that slide?"
The lesson of history for Israel is that any or all of their neighbours may attack them, or be used for proxy wars against them, but when they're attacked it does in fact work to hit back brutally and without mercy in a way that will remind people not to antagonise them for a couple of decades at least. That's why the Golan Heights are now part of Israel, whether the international law weenies recognise it or not.
Why does the US let that slide? You try telling the people of Israel that having 1,000 mostly civilians raped, murdered, mutilated or abducted by bearded scumbags in a couple of days is something they should just suck up. For comparison, the national casualty rate is equivalent to around 8,000 people in an attack on the UK, or 30,000 in an attack on the US.
I also note a hint of "victim blaming" by suggesting it's somehow Israel's fault for being more trusting of the people of Gaza. Gazans chose to elected Hamas, Hamas chose violence, and perhaps they might reflect on that. I'd agree Netanahyu is (a) a complete arsehole, (b) a crook, and (c) has big questions to answer to his own country, but that still doesn't make him or Israel to blame.
I really am no fan of Israel, but if Hamas didn't want this outcome maybe they shouldn't have started it.
History isn't likely to look back so kindly on the era of foolish belligerence embodied by Netanyahoo starting fights he expected the US to finish if they got of hand. It won't look kindly at the US for allowing it's leadership to zombie walk us into this horror show chanting "must be a friend to Israel" as if backing ultra-zionists was ever going to lead to peace in the region.
The mass attack of Hamas on civilian was barbaric, and a war crime. The response should have been to hunt those responsible to the end of the earth and the end of time, like the war criminals of the past. Instead the current regime used it as an excuse to conduct genocide of a magnitude far beyond the tragedy they suffered. Also let's not white wash the fact that Net had been provoking the Palestinians for months trying to incite a response big enough to justify a crackdown, all in the service of creating a distraction from Net's corruption charges and failing coalition of leadership.
He provoked a more massive response than he anticipated, ignored all the warnings of the coming attack, and relied on an ineffective Israeli version of the "Great wall of Trump" instead of securing the border with adequate troops. Hamas is still culpable for their part in the war crimes, and even the context of the outright provocation of the attack don't justify attacking civilians, but that doesn't wash the blood from Israel's hands either.
So Hamas fighters will be hunted for decades wherever they try to hide around the world, but is the senseless decimation of Gazan civilians also worth condemning a generation of young Israeli soldiers to the same fate? Because the crimes the world are watching have no expiration. Many of those that fight in this war may never be able to leave Israel again without the risk of arrest. And while the current US president may never waver in his choice to refuse to hold Netanyahu accountable and make the nation complicit in genocide, future administrations may not.
Israel should look to the likely future, one where it's brash and monstrous actions have cost it the allies the hold the larger wolves in the region at bay. If it continues to act as a terrorist state itself, it may find itself called to account, shut off from the world, and left to contend with the neighbors it has been bullying for generations on it's own.
Dropping a 2000lb bomb on a building full of civilians because of a possible report of a suspected Hamas combatant isn't a fight it's a fig leaf for a massacre. Cold numbers show that deaths Israel inflicted vastly exceed what they have suffered, and that they are ignoring the methods and norms by which they themselves operated in previous conflicts in Gaza. They, like the US did have lost their way and lost their minds. They are failing in their declared objective of eliminating Hamas, while sowing the wind ensuring the maimed children that survive will likely take up the cause and return to decades of suicide bombing.
Rooting out Hamas without indiscriminately targeting the rest of the Palestinians would have taken time and come at a high cost, but it was the only legal option. Hamas's crimes don't justify Israels, and as they wield absolute power over Gaza they have no excuse for abandoning what few bounds which are supposed to be enforced in war. Bombing hospitals instead of clearing them and allowing them to continue to treat the wounded. Targeting all civilian infrastructure like water, sewage and power. Blockading basic food and medicine.
That's not a fight at at, and trying to spin it as one isn't making your argument more convincing.
Many years ago, around the time of the Google-AI-drone drama, an military spokesperson was promoting AI for drones saying (paraphrased from memory) that AI would distinguish between men carrying hoes and men carrying guns, and thus save lives. Which totally ignores everything from the fact that at that time in Afghanistan most men carry guns, and someone carrying a hoe could be hiding a gun, or a roadside mine, etc., all the way to asking "what is the US end goal in Afghanistan?".
So how will this AI "training" work get evaluated for quality? Sorry, that's confidential. Broadly speaking, US wars in the ME and Afghanistan have been fighting opponents that are guaranteed to not be able to win a battle. The current conflict included. One side effect of that is not learning through experience anything about fighting an opponent who CAN win a battle (and is watching now, and knows it).
Why is the man and his group who is responsible for not having the 2K soldiers posted on the Gaza border, despite repeated intelligence warnings from border observation posts that Hamas was training for an incursion, now in charge of fucking further up in Gaza and taking every opportunity to further inflame passions in the West Bank too? How is that in the US interest?
FWIW - I don't support BDS or the abolition of Israel, and see Intel's multi billion investment in a plant in Israel as positive thing. All the more reason to prioritize peace and stability over expansion and revenge. The current strategy is Gaza is not the only one possible. At a bare minimum, Gazans could have had the opportunity to leave Gaza for refugees camps in Israel, with access to food and medical care.
What does any of that have to with the current Houti aggression? If you can't see that AI won't help you anyway.
"Every step that involves AI has a human checking in at the end."
...because in a war zone the AI is really bad at recognising your own team, whereas a human is merely unreliable.
(In most modern conflicts, the hi-tech weaponry is only on one side, leading to the exacerbating factor that only one side is actively trying to hide from the AI.)
For AI, humans are a problem that needs to be eliminated in any serious questions about AI functionality.
Oh wait I need to be registered as saying, "For humans AI is a problem that needs to be eliminated in any serious questions about AI functionality."
Oh dear, I've just been deleted after AI processed this post.
A lot of this AI stuff is automating stuff that humans used to do manually, and is quite well understood.
Want to know where the Russian fuel depot is in Ukraine? Simply look at ground radar records and track all the movements of trucks - then work out where they're going to fill up and distribute to the forward units. This also works for ammunition depots. It used to be done by recording the data onto videotape and fast-forwarding re-winding and looking for patterns.
In Yemen we're looking for missile launch sites, that can be easily camouflaged. But the missiles and fuel have still got to get transported to them. So we can try to track those. Or there may be only a few expert firing teams, in which case you might be able to track them by radio / mobile phone intercepts or human intelligence or whatever. So you build a bunch of metrics and chuck that at masses of sensor data from all your various intel assets and see what comes out.
I imagine they're also trying to use AI to generate patterns from the data that hadn't previously been thought of. I'd be scepitical, but you never know.
Intelligence has had the problem of too much data, and too few analysts since at least WWII. That was one of the main secrets of Bletchley Park, sure they were amazing at codebreaking - but that would have been a lot less useful if they hadn't also been amazing at automating the process.