【Hacker News搬运】五角大楼的硅谷问题

[hackernews]

Title: The Pentagon's Silicon Valley Problem

五角大楼的硅谷问题

Text:

hn link

Url: https://harpers.org/archive/2024/03/the-pentagons-silicon-valley-problem-andrew-cockburn/

该文章讨论了五角大楼在采用和实施先进人工智能技术时所面临的挑战，以及硅谷和美国军队之间的关系。文章指出，以色列的基于人工智能的安全系统在预测哈马斯的袭击方面失败了，并且人工智能在理解人类意图方面存在局限。文章还批评了五角大楼的官僚主义阻碍了人工智能的采用，尽管五角大楼资助了许多人工智能项目。文章认为，将软件视为解决人类冲突的解决方案是一个神话，而且军事工业复合体对塑造科技行业的影響力大于反之。作者还强调了在军事项目中工作的科技员工所面临的道德问题，以及科技公司与国防部门之间的持续合作。

该文章讨论了科技公司日益参与国防产业的趋势，尽管像谷歌这样的公司最初表现出了犹豫。退出Maven合同引起了愤怒，但现在亚马逊、微软、甲骨文和谷歌等公司正在争夺政府合同。文章提到了在乌克兰战争中无人机的使用，双方都有效地使用了简单、便宜的设备。科技公司急于展示他们的能力，Anduril宣布开发了一种能够自主检测和摧毁威胁的无人机。文章还讨论了尽管对其有效性存在担忧，Anduril仍获得了一份价值10亿美元的合同，用于开发一种反无人机系统。文章还提到了五角大楼对人工智能技术的投资，以及使用人工智能在战争中可能存在的伦理问题。

Post by: NDAjam

Comments:

neonate: <a href="http://web.archive.org/web/20240327152111/https://harpers.org/archive/2024/03/the-pentagons-silicon-valley-problem-andrew-cockburn/" rel="nofollow">http://web.archive.org/web/20240327152111/https://harpers.or...</a>

neonate: <a href=“http://x2F；&x2F；web.archive.org&#x2F；web&#x2F：20240327152111&#x2F”https:&#x2F！&#x2F，harpers.org&#x2F；archive&#x2F…2024&#x2F？03&#x2F。五边形硅谷问题andrew cockburn&#x2F“rel=”nofollow“>http://x2F&#x2F；web.archive.org&#x2F；web；20240327152111；https:&#x2F&#x2F；竖琴。或者</一

dosinga: The examples in the article are rather cherry-picked. Failures in Vietnam can hardly be blamed on an IBM 360 only. The Hamas attack might have surprised Israel but the Iron Dome has been tech working well in recent years. The US warned anybody who wanted to listen (not many) that Russia was about to attack Ukraine. And it was a bunch of rather theoretical physicists who built the atomic bomb.

dosinga: 文章中的例子是精心挑选的。越南的失败很难仅仅归咎于IBM 360。哈马斯的袭击可能会让以色列感到惊讶，但近年来铁穹的技术一直运行良好。美国警告任何想听的人（不多），俄罗斯即将袭击乌克兰。是一群相当理论化的物理学家制造了原子弹。

Barrin92: It's just the usual technology obsession of military industrial and political types that's been around for decades. The reality is that the most important factor in combat is the human one and every fancy gadget you use just introduces more liability and weak points.<p>The AI marketing hype and lobbying stuff fills the pockets of a few people but it doesn't make soldiers more effective, "cloud computing controls the battlefield" is such a meme worthy sentence I don't understand how anyone can take someone seriously who says that out loud.<p>What you could see in the Israel-Hamas conflict mentioned in the article is what you also see with the Houthis or in Ukraine, that the best technology on the battlefield is cheap, resilient and simple enough to be understood and operated by the least competent soldier, not some 10 billion dollar fantasy tool out of a sci-fi novel.<p>The example in the article of Hamas feeding Israeli informants deliberate misinformation to strengthen the notion that Hamas would not attack, now imagine this amplified by even more gullible LLM powered "intelligence analysts". It's a theme of the "AI age", the people who stand to benefit the most are critically thinking humans able to exploit the tool induced stupidity of everyone else. Hackers, appropriately enough.

Barrin92: 它；这只是军工和政治类型通常对技术的痴迷；它已经存在了几十年。事实是，战斗中最重要的因素是人，你使用的每一个花哨的小工具都会带来更多的责任和弱点<p> 人工智能营销炒作和游说充斥着少数人的口袋，但事实并非如此；t使士兵更有效率，&quot；云计算控制战场”；是这样一个值得模因的句子吗；I don’我不明白怎么会有人把大声说这话的人当回事<p> 你可以在文章中提到的以色列-哈马斯冲突中看到，你也可以在胡塞人或乌克兰看到，战场上最好的技术是廉价、有弹性和简单的，足以让最不称职的士兵理解和操作，而不是科幻小说中价值100亿美元的幻想工具<p> 文章中的例子是哈马斯向以色列线人提供故意的错误信息，以加强哈马斯不会发动袭击的观念，现在想象一下，这被更容易上当受骗的LLM所放大；情报分析员”；。它；这是“；AI年龄”；，受益最大的人是具有批判性思维的人，他们能够利用其他人因工具而产生的愚蠢。黑客，足够恰当了。

nkozyra: I think - like a lot of media reporting on the space - this overgeneralizes (heh) artificial intelligence. The predictive aspects of ML have been in use in modern militaries for decades, and the opening graf handwavely indicates that an LLM was a bigger chunk of the perceived intelligence failure of the October 7 attack.<p>That an LLM is a part of a system that includes a large amount of ML is not surprising. It's a great human interface. Do I for a second believe that it played a much larger role, such to be implied as responsible in any non-negligble way for missing the attack. Of course not.<p>My point here is that ML continues to play a role, ML continues to both succeed and fail, and ML will continue to be imperfect, even moreso as it competes against adversarial ML. Blaming imperfect tools for inevitable failures is not a useful exercise, and certainly not a "problem" considering the alternative being even more failure-prone humans.

nkozyra: 我认为，就像许多媒体对太空的报道一样，这过度概括了人工智能。ML的预测方面已经在现代军队中使用了_decades_，开场白手工表明，LLM在10月7日袭击的感知情报失败中占了更大的比例<p> LLM是包含大量ML的系统的一部分，这并不奇怪。它；这是一个很棒的人机界面。我是否认为它发挥了更大的作用，暗示它对错过袭击负有任何不可忽视的责任。当然不是<p> 我在这里的观点是，ML继续发挥作用，ML继续成功和失败，ML将继续是不完美的，甚至在与对抗性ML竞争时更是如此；问题“；考虑到另一种选择是更容易失败的人类。

Const-me: I think the most important lesson, it’s borderline impossible to design any good system without clear use cases.<p>Ukraine has these use cases, also high motivation to tackle them. Ukrainians are controlling battlefield with commodity computers <a href="https://en.defence-ua.com/news/how_the_kropyva_combat_control_system_helps_in_the_most_difficult_situations_fortified_positions_couldnt_save_russian_army-3646.html" rel="nofollow">https://en.defence-ua.com/news/how_the_kropyva_combat_contro...</a> They sunk multiple Russian warships with long-range naval drones <a href="https://www.bbc.com/news/world-europe-68528761" rel="nofollow">https://www.bbc.com/news/world-europe-68528761</a> They recently started large-scale testing of cheap flying drones with computer vision-based target recognition on board <a href="https://www.forbes.com/sites/davidhambling/2024/03/21/ukraine-rolls-out-target-seeking-terminator-drones/" rel="nofollow">https://www.forbes.com/sites/davidhambling/2024/03/21/ukrain...</a><p>However, US is at peace. Which is a great thing by itself, but it means it’s too easy for them to waste billions of dollars developing technologies which look awesome in PowerPoint, but useless in practice.

Const-me: 我认为最重要的教训是，如果没有明确的用例，设计任何一个好的系统都是不可能的<p> 乌克兰有这些用例，也有很高的动机来解决它们。乌克兰人正在用商品计算机控制战场<a href=“https://；&#x2F；en.dedefense ua.com&#x2F！news&#x2F：how_the_kropyva_combat_control_system_help.in_the_most_dffice_situations_focertified_positions_couldnt_save_russian_army-3646.html”rel=“nofollow”>https://&#x2F；en-defense ua.com&#x2F；news&#x2F；如何控制</a> 他们用远程海军无人机击沉了多艘俄罗斯军舰&#x2F；www.bbc.com&#x2F；news&#x2F；world-europe68528761</a>他们最近开始对廉价飞行无人机进行大规模测试，并在飞机上进行基于计算机视觉的目标识别<a href=“https://；&#x2F；www.forbes.com&#x2F”sites&#x2F，davidhalling&#x2F2024&#x2F）03&#x2F！21&#x2F&#x2F；www.forbes.com&#x2F；站点；davidhalling；2024；03；21；乌克兰</a> <p>然而，美国是和平的。这本身就是一件好事，但这意味着他们很容易浪费数十亿美元开发在PowerPoint中看起来很棒但在实践中毫无用处的技术。