EPTapnews.com
Things Aren’t Peaceful in Palestine - Page 250
| Forum Index > General Forum |
NOTE: When providing a source, please provide a very brief summary on what it's about and what purpose it adds to the discussion. The supporting statement should clearly explain why the subject is relevant and needs to be discussed. Please follow this rule especially for tweets. Your supporting statement should always come BEFORE you provide the source. | ||
|
Cricketer12
United States13974 Posts
apnews.com | ||
|
JimmiC
Canada22817 Posts
| ||
|
JimmiC
Canada22817 Posts
| ||
|
Nebuchad
Switzerland12202 Posts
I like this quote a lot because it highlights everything wrong with reliance on machine learning. The machine didn't do it "coldly", because the machine didn't do it at all. Someone coded the machine to learn the things it has learned in the way that it has learned them. If you had someone else code the machine, maybe someone who wasn't part of a fascist government that is trying to settle land occupied by a subhuman population, the code would have been different, and the results would have been different. But because the person ordering the deaths went through the hassle of machine learning instead of just ordering the deaths directly, even to this guy who is providing commentary against the system it still looks like an objective, almost superhuman way of reaching a decision. | ||
|
Jockmcplop
United Kingdom9653 Posts
On April 04 2024 00:00 JimmiC wrote: I don’t think those numbers would be common. Israel says it is 1-1 on combatants to civilians and other things I’ve read is 5-1. Now I’m not saying those don’t happen just that I don’t think 15,20-1 is common or that would mean a ton of attacks just kill Hamas and that seems pretty unbelievable as well. The problem isn't that Israel have killed 15-20 civilians for every Hamas fighter, its that they've set that as a guideline or limit. Imagine if they did kill that many innocent civilians. With a target list of 37,000 set by the AI that's at least 555,000 dead. That's what the people making these decisions deem reasonable. | ||
|
JimmiC
Canada22817 Posts
| ||
|
Nebuchad
Switzerland12202 Posts
On April 04 2024 03:18 Jockmcplop wrote: The problem isn't that Israel have killed 15-20 civilians for every Hamas fighter, its that they've set that as a guideline or limit. Imagine if they did kill that many innocent civilians. With a target list of 37,000 set by the AI that's at least 555,000 dead. That's what the people making these decisions deem reasonable. 37000 is also quite a large number for Hamas membership, estimates I've seen were around 25000, so the fact that this machine spits that number is probably indicative of a very low standard used to determine what's an acceptable target. | ||
|
Jockmcplop
United Kingdom9653 Posts
On April 04 2024 03:50 Nebuchad wrote: 37000 is also quite a large number for Hamas membership, estimates I've seen were around 25000, so the fact that this machine spits that number is probably indicative of a very low standard used to determine what's an acceptable target. Yeah for sure. From what we've seen I would think the target list is completely inaccurate in the majority of cases. For all the numbers we were given the expected accuracy of the AI wasn't one of them. | ||
|
hitthat
Poland2267 Posts
| ||
|
Magic Powers
Austria4149 Posts
I don't know how much longer they expect to be able to get away with this lie. I'm sure plenty of people believe it, but it resembles Russian-style propaganda at this point. https://www.reuters.com/world/middle-east/gaza-war-after-six-months-what-are-issues-now-2024-04-04/ | ||
|
Gorsameth
Netherlands21699 Posts
| ||
|
Nebuchad
Switzerland12202 Posts
https://www.972mag.com/lavender-ai-israeli-army-gaza/ | ||
|
r00ty
Germany1056 Posts
On April 04 2024 20:53 Nebuchad wrote: I recommend reading the 972 article that was the basis for the Guardian article of one page ago, some of the details in there are worth being aware of. https://www.972mag.com/lavender-ai-israeli-army-gaza/ Brave new world. Getting rid of the human bottleneck, that's a way to put it i guess. I think it's absolutely horrifying and a crime against humanity. Dear god what are we doing? Are there really people who think using that AI is a good idea? | ||
|
Acrofales
Spain18004 Posts
On April 04 2024 21:21 r00ty wrote: Brave new world. Getting rid of the human bottleneck, that's a way to put it i guess. I think it's absolutely horrifying and a crime against humanity. Dear god what are we doing? Are there really people who think using that AI is a good idea? I think using AI is a good idea for some things and a bad idea for other things. I even think using AI for targeting missiles could be a good thing. Not as if humans are infallible either. How many weddings did the US drone strike in Iraq? That said, the key here is accountability. In the case of Israeli bombing campaigns everybody is washing their hands of the innocent deaths this massive bombing campaign has cost. The AI sure as hell isn't accountable. It's just doing what was programmed. The programmers aren't accountable, they just put in the parameters they were given. The military brass isn't accountable, those numbers came from the legal department on "proportionality", etc. etc. | ||
|
r00ty
Germany1056 Posts
On April 04 2024 21:21 Acrofales wrote: I think using AI is a good idea for some things and a bad idea for other things. I even think using AI for targeting missiles could be a good thing. Not as if humans are infallible either. How many weddings did the US drone strike in Iraq? I meant that "Lavender" AI specifically. Also not looking forward to my job getting replaced by another AI, but that's a different topic. That said, the key here is accountability. In the case of Israeli bombing campaigns everybody is washing their hands of the innocent deaths this massive bombing campaign has cost. The AI sure as hell isn't accountable. It's just doing what was programmed. The programmers aren't accountable, they just put in the parameters they were given. The military brass isn't accountable, those numbers came from the legal department on "proportionality", etc. etc. Yeah, that's the scary part. | ||
|
Dan HH
Romania9122 Posts
On April 04 2024 21:21 Acrofales wrote: I think using AI is a good idea for some things and a bad idea for other things. I even think using AI for targeting missiles could be a good thing. Not as if humans are infallible either. How many weddings did the US drone strike in Iraq? That said, the key here is accountability. In the case of Israeli bombing campaigns everybody is washing their hands of the innocent deaths this massive bombing campaign has cost. The AI sure as hell isn't accountable. It's just doing what was programmed. The programmers aren't accountable, they just put in the parameters they were given. The military brass isn't accountable, those numbers came from the legal department on "proportionality", etc. etc. I don't think this is a game-changer for accountability. Technically, the system only recommends targets, the strikes still have to be human approved. If they treat that approval as a formality, that's on them. And even for target recommendation, the AI isn't setting its own tolerance thresholds, that's an executive decision. They can probably turn those 37k targets to 5k or 50k with the flick of a slider. | ||
|
Magic Powers
Austria4149 Posts
To illustrate with numbers from 0-100: A) Operator slides to 0-20, AI achieves 5 (demonstrating that the AI can't bring casualties below 5). B) Operator slides to 0-80, AI achieves 5 (effectively equal to A, demonstrating that the AI works as intended). C) Operator slides to 20-50, AI achieves 20 (demonstrating that the AI still does its job). D) Operator slides to 60-80, AI achieves 60 (AI still does its job). There's nothing wrong with the AI. A problem can only arise when humans direct the AI (due to a meta goal such as destroying Hamas or forcing them to surrender, which is not the job of the AI). Here's how it happens: Assuming that neither A, B nor C can accomplish the set military goal (win the war; destroy Hamas), but D can accomplish it (because the military goal requires more casualties to be realistic), then the operator is incentivized to set to D, and therefore not to minimize casualties. If he sets to C, then the AI may be "too efficient" at minimizing casualties, thus resulting in a failure of the meta goal (to win the war). This is because AI can't figure out how to accomplish the meta/military goal while minimizing casualties, that task would be far too complex. Such an AI can't possibly exist, it'd give us literally godlike knowledge of the universe. AI can't figure out how many casualties are required, it can only figure out how to minimize casualties within the set parameters anywhere between 0-100. | ||
|
JimmiC
Canada22817 Posts
| ||
|
GreenHorizons
United States23245 Posts
On April 04 2024 21:21 Acrofales wrote: I think using AI is a good idea for some things and a bad idea for other things. I even think using AI for targeting missiles could be a good thing. Not as if humans are infallible either. How many weddings did the US drone strike in Iraq? That said, the key here is accountability. In the case of Israeli bombing campaigns everybody is washing their hands of the innocent deaths this massive bombing campaign has cost. The AI sure as hell isn't accountable. It's just doing what was programmed. The programmers aren't accountable, they just put in the parameters they were given. The military brass isn't accountable, those numbers came from the legal department on "proportionality", etc. etc. Last year, in relation to Israel killing a Reuters journalist, I asked about what the supposed "investigations" by Israel amount to when it comes to accountability On December 09 2023 03:10 GreenHorizons wrote: A Reuters investigation has concluded that Israel killed one of their journalists in Lebanon and Israel either knew or should have known they were targeting and firing on journalists. www.npr.org As a bit of an aside, I always read "The Israeli military said after the attack that it was investigating the incident." but I rarely if ever come across "after the investigation they were charged and convicted of X" Does anyone have any data/statistics on what comes of these alleged "investigations"? Sure smells a lot like the old US police refrain of "we investigated ourselves and found we did nothing wrong". So, did anyone ever find the data/statistics (even from before Oct 7 would be something) on what comes of these supposed "investigations" Israel does regarding accountability? I think we might have better odds wishing on a star (that turns out to just be a satellite) than relying on Israel's investigations for accountability. | ||
|
JimmiC
Canada22817 Posts
| ||
| ||
