What aspects of modern warfare didn't Hideo Kojima foresee?
>Another combat veteran, now with a Pentagon agency working on these issues, told me that the AI developers he works with didn’t seem to understand some of the requirements for the technology’s military application. “I don’t know if AI, or the sensors that feed it for that matter, will ever be capable of spontaneity or recognizing spontaneity,” he said. He cited a DARPA experiment in which a squad of Marines defeated an AI-governed robot that had been trained to detect them simply by altering their physical profiles. Two walked inside a large cardboard box. Others somersaulted. One wore the branches of a fir tree. All were able to approach over open ground and touch the robot without detection.
Oh..
>I was curious about Palantir, whose stock indeed soared amid the 2023 AI frenzy. I had been told that the Israeli security sector’s AI systems might rely on Palantir’s technology. Furthermore, Shin Bet’s humiliating failure to predict the Hamas assault had not blunted the Israeli Defense Force’s appetite for the technology; the unceasing rain of bombs upon densely packed Gaza neighborhoods, according to a well-sourced report by Israeli reporter Yuval Abraham in +972 Magazine, was in fact partly controlled by an AI target-creation platform called the Gospel. The Gospel produces automatic recommendations for where to strike based on what the technology identifies as being connected with Hamas, such as the private home of a suspected rank-and-file member of the organization. It also calculates how many civilians, including women and children, would die in the process—which, as of this writing, amounted to at least twenty-two thousand people, some 70 percent of them women and children. One of Abraham’s intelligence sources termed the technology a “mass assassination factory.” Despite the high-tech gloss on the massacre, the result has been no different than the slaughter inflicted, with comparatively more primitive means, against Dresden and Tokyo during World War II.
> at least twenty-two thousand people, some 70 percent of them women and children.
People are definitely dying and that includes civilians, but facts matter and a lot of these numbers are simply made up by Hamas. Here's some analysis that demonstrates that it's extremely unlikely Hamas's Gaza Health Ministry numbers are based in reality: https://www.tabletmag.com/sections/news/articles/how-gaza-he...
Stuff like almost perfectly linear growth of deaths over days and so on, take a look at the article.
This is also Hamas who claims that every person killed was a civilian. Try to find "number of Hamas combatants killed" in the Gaza Health Ministry numbers, they don't even count this. Isn't that a bit weird? Israel will tell you how many people killed were armed soldiers and how many were civilians, as most places will do.
I don't know why people believe the numbers coming from Hamas. I wish there were a more reliable source in the area, but believing Hamas because you have no other numbers to go on is plain stupid.
The thing to do when you feel like a commenter is "beyond reason" is to flag the comment. Writing a long response to what you feel is an unreasonable or bad-faith comment is just feeding the toxicity. The abyss is commenting back into you, so to speak.
You can see the effect here: in trying to compose a reply to the parent comment, you managed to violate a variety of HN guidelines yourself.
Ah OK. I didn't realize I was violating guidelines. I'll try flagging in the future.
I thought there was some utility to analyzing the comment because it's characteristic of much of the rhetoric on this topic and it seemed like a good "teachable moment." But point taken.
I’d like to note that it is not Hamas which is reporting the official death count, it is the Gaza Health Ministry. While the ministry is under the control of Hamas, the actual reporting is done by actual public health professionals, doctors, etc. who may be Hamas members simply as means of occupation. So Hamas is not under reporting anything.
Second, the methodology of the reporting is a lower bound. They compile a list of names who health-care workers (doctors, nurses, etc.) confirm dead. This list is very credible, as you can cross reference the names with the Israeli controlled Gaza population registry, you can find real people behind those names, with social media profiles, obituaries, etc. This methodology has been proven to be accurate in previous wars. This is still a lower bound, because there are still around 7000 people missing and presumed dead. This number is harder to estimate because this may actually include double counting, and false positives (e.g. successfully fled over the border). It may also be an under count as there are cases of entire families being bombed at the same time with nobody left to report them missing. Your parent’s 100 000 is not realistic, but around 40 000 is not an unrealistic estimate (the 100 000 is used to described the sum of dead + missing + injured, so perhaps this was a simple mistake by your parent).
And finally the statistical methodology of the article you cited earlier is—to but it bluntly—pseudostatistical crap. The your parent was correct in pointing out that the assumptions based in that article are never sufficiently justified to proof an abnormality. This is actually from the playbook of Trump election denialism, where the distribution of votes was somehow proof of tampering. Here the distribution of reported death is not proof of any abnormality. This article you cited has actually been submitted and flagged, submitted and flagged repeatedly here on HN. The HN community has weighted that article and deemed it unfitting. (https://news.ycombinator.com/item?id=39684474)
Why would that be the case. If anything, the last 20 years have reinforced the idea that if enemy combatants simply embed themselves in civilian populations, they are virtually impossible to target without mass collateral damage.
No no no no no no no, please no. Under no circumstances is that OK.
If the enemy is in a group of civilians, you use the police, arrest them, put them for a trial and send to prison.
If the enemy is in a group of civilians, and your immediate action is to bomb them along with the civilians, then you don’t value the lives of those civilians. At best you don’t care whether those civilians are alive or dead, and at worst you actually want those civilians dead. In the former case, you are doing a war crime, a crime against humanity, in the latter case you are doing a genocide and using the enemy as an excuse or justification.
In any case, there are plenty of ways to get around that. Unless the mass casualty is the point.
How many places in a densely populated city can you precisely drop a bomb with no risk to civilians? Yes, technology has improved, but my point is that better tech doesn't just automatically make war less awful.
I don't believe the people operating the weapons are different in ways that make civilian casualties obsolete. In WW2 it was a decision to bomb population centers, not an accident. That decision can be made today too.
Even 100% accurate precision weapons are only as good as the target information supplied to them. Bad Intel and misidentified targets are very real factors that can cause a precision weapon to strike something they shouldn't.
People are so desperately wanting to believe that AI will give them the revealed truth. Such systems should be named "Racist Uncle Dave" because they hallucinate some answer everytime they open their virtual mouths with some probability of being somewhat correct this time.
I’ve been hanging around some MBA types lately, and I’m coming to realize the product doesn’t actually have to perform to high specs like engineers would demand. Palantir is selling a “story” that their AI system magically decreases casualties and finds good targets, and it’s got firm numbers printed in the console (that could be completely wrong) but that is more convincing to MBA’s than any wishy intuition. So the MBA’s buy into the marketing, and the executors are buying into offloading their conscience.
please. the last people that would buy into any kind of marketing are MBAs because they actually study marketing among other things during their 2 years of obtaining an MBA.
there is no perfect product, I don't know how well Palantir AI works, but I would be surprised it doesn't work at all
McKinsey hires top of the class people at top schools and pays them top dollar. If all they needed was marketing or sales, they could go to University of Nowhere and hire good looking tall people to do sales.
It's not hard because of cognitive ability required.
It's hard because it's expensive fluff bs for jobs that have it as requirement. Useless without a position where you can exploit having MBA diploma.
There's a case in Poland that one "university" was giving MBAs like candies to party officers so they can have some higher education to occupy well paid leadership positions in publicly owned companies.
Nobody said MBAs buying Palantir or whatever came from "top" MBA programs. In Poland they would come from this "university".
Regardless, I don't think you can make a reasonable case that MBAs in general are such low IQ people that they are buying Palantir software for no reason other than marketing or sales. It's much more likely that Palantir software actually works at least somewhat well. Palantir has a 50 billion dollar market cap. You don't get to that level on marketing alone
Overselling supposedly magic software is not a kind of fraud?
I said that "I'm sure" that Palantir has a bit more than Bernie because they actually produced some software, I believe.
"Not by much" was my guess given my experience of living 45 years on this planet through AI winter and spring (multiple?), working with software and computers. Experiencing how much of marketing is straight up lies. Seeing firsthand how government spends money on future predicting software that doesn't actually accurately predict anything, just presents results of some modelling.
But that's just my guess. Maybe it is magic this time. After all my expectations were contradicted once already with ChatGPT. Not strongly, but somewhat.
"Overselling supposedly magic software is not a kind of fraud?". I suggest googling what fraud means. Marketing is most definitely not fraud and has nothing to do with fraud. You are making circular arguments. Palantir is a 50 bln dollar company and you provided exactly zero evidence that they are "overselling supposedly magic software"
in other words you have no idea what you are talking about specifically with respect to Palantir. I never understand why people do this. Why make these strong claims if it turns out you just don't know ?
Sorry, were you expecting to obtain unassailable secret evidence against 50bln dollar scheme from a random commenter on the internet? Apologies for not measuring up to your expectations. Please notify me once you obtain unassailable evidence that this scheme is in fact not a (colloquially) fraud i.e. it's worth what was spent on it.
I wonder if they were a part of decision making process that caused Isreal to methodically bomb Central World Kitchen aid convoy manned with foreign volunteers. We'll never know because if they were this information won't be publicized because of relationships between Isreal, US and business.
When you read up on Palantir budiness in Israel there are question being raised why money is funneled back to a US corporation instead of developing local Isreali competence in processing intel.
This is horrible news. Blurring the lines of accountability between people and software in the industry of war is a recipe for Armageddon. It's not only genocide laundering, which is atrocious enough. Unchecked, it will lead to a "stand your ground" type of situation where countries may strike first in anticipation of other actions. I fear for the future.
> according to a well-sourced report by Israeli reporter Yuval Abraham in +972 Magazine, was in fact partly controlled by an AI target-creation platform called the Gospel
Don't try to "fake-news" it because it doesn't fit with your narrative.
Tech is being abused and combined with already authoritarian-fascist policies, is killing civilians en masse.
The point remains: software is continually used as the scapegoat when things go wrong to shield human actors. Practitioners in our, admittedly young, field have shown very little appetite for taking any sort of responsibility expected of engineering professionals who inflict harm.
No one is allowed to attack other users like that, regardless of how right they are or feel they are.
If you don't want to be banned, you're welcome to email hn@ycombinator.com and give us reason to believe that you'll follow the rules in the future. They're here: https://news.ycombinator.com/newsguidelines.html.
>Another combat veteran, now with a Pentagon agency working on these issues, told me that the AI developers he works with didn’t seem to understand some of the requirements for the technology’s military application. “I don’t know if AI, or the sensors that feed it for that matter, will ever be capable of spontaneity or recognizing spontaneity,” he said. He cited a DARPA experiment in which a squad of Marines defeated an AI-governed robot that had been trained to detect them simply by altering their physical profiles. Two walked inside a large cardboard box. Others somersaulted. One wore the branches of a fir tree. All were able to approach over open ground and touch the robot without detection.
Oh..
>I was curious about Palantir, whose stock indeed soared amid the 2023 AI frenzy. I had been told that the Israeli security sector’s AI systems might rely on Palantir’s technology. Furthermore, Shin Bet’s humiliating failure to predict the Hamas assault had not blunted the Israeli Defense Force’s appetite for the technology; the unceasing rain of bombs upon densely packed Gaza neighborhoods, according to a well-sourced report by Israeli reporter Yuval Abraham in +972 Magazine, was in fact partly controlled by an AI target-creation platform called the Gospel. The Gospel produces automatic recommendations for where to strike based on what the technology identifies as being connected with Hamas, such as the private home of a suspected rank-and-file member of the organization. It also calculates how many civilians, including women and children, would die in the process—which, as of this writing, amounted to at least twenty-two thousand people, some 70 percent of them women and children. One of Abraham’s intelligence sources termed the technology a “mass assassination factory.” Despite the high-tech gloss on the massacre, the result has been no different than the slaughter inflicted, with comparatively more primitive means, against Dresden and Tokyo during World War II.