互联网与文化antipope.org
老兄，你毁了未来
Dude, you broke the future
2018 — 2020，它一次次回到首页
黑点是上首页（≥ 50 分）的那次投稿，空心点是没冲上去的投稿，点越大分数越高。
导读
科幻作家查理·斯特罗斯（Charlie Stross）2018年的演讲文章，提出一个核心观点：公司本身就是一种“古老、缓慢的AI”——无道德、目标单一、为生存而不断优化的实体。作者借此论证，我们连这些“慢AI”（如社交媒体成瘾、精准广告、虚假信息）都管不好，凭什么相信未来能控制真正的人工智能？文章还讨论了1970年代以来反垄断监管失效、科技巨头失控等问题。
为什么经久不衰：它把“AI风险”从遥远的科幻拉回到我们已被公司、广告和算法支配的当下，提供了一个至今仍适用的思考框架：先学会管好现有的优化机器，才谈得上管好未来的AI。
HN 评论精选
公司因云基础设施而变得敏捷且能无限扩张，工业文明真正起步不过两百年
我很同意演讲中的很多观点。真正让我们头疼的是公司越来越擅长最大化股东价值。过去大公司相对静态，进入某个行业、建好工厂就待在那里；现在一些公司能敏捷地转向新业务并接管市场。所有“科技”公司用的都是差不多可以从亚马逊租来的基础设施，所以敏捷变得很便宜。另外，公司现在可以做得更大——重工业有规模瓶颈，通用汽车几十年前就因太大而尾大不掉，但谷歌、苹果、亚马逊、沃尔玛似乎没有这个问题。工业文明其实比斯特罗斯说的还年轻，不到200年：1830年利物浦—曼彻斯特铁路开通，普通人第一次能买票坐定时列车，那才是工业革命“出测试版”开始规模化的时刻。
Animats查看英文原文
I agree with a lot of that talk. What's getting us now is that corporations are getting better at maximizing shareholder value. For a long time, big companies were relatively static entities. They got into some line of business, built the necessary factories or railroad tracks or stores, and mostly stayed there. Now, some of them pivot into new businesses with agility and take over. All the "tech" businesses use roughly the same infrastructure, which you can rent from Amazon. So agility is cheap. Another point is that companies can be bigger now. There are scaling problems with heavy industry. General Motors ran into scaling problems decades ago - they were too big to get out of their own way, and management couldn't keep track of what was happening in distant plants. Google and Apple don't seem to have that problem. Neither does Maersk, the world's largest shipping line. Or DHL or FedEx or UPS or Amazon or WalMart or McDonalds. Industrial civilization is even younger than Stross says. Less than 200 years. A good start date is the opening of the Liverpool and Manchester Railway Railway, 1830. For the first time, anyone could buy a ticket on a scheduled train and go someplace. This is when the industrial revolution got out of beta and started to scale. There were steam engines for a century before that, but they were isolated one-offs. The Manchester and Liverpool had many engines, double track, stations, timetables, signals, and paying customers. That's when things really started to move. Quite literally.
把公司比作AI还不够——军队才是顶级掠食者AI，公司只是食草动物
“公司已经是AI，而且是最糟糕的那种（多个、互相争斗、改造环境、被生存驱动）”这个观察很有意思。我认为还有一类顶级掠食者AI：军队。公司在这点上远不如军队——公司AI还得乖乖向军队纳税，军队用这些资源开发和限制最危险的技术。我们已经打过两场顶级AI战争，还有一场差点毁灭一切的冷战AI战争，救了我们的正是顶级AI的生存本能。公司只是第二梯队，是这个生态里的食草动物，而我们是被吃的植物。比起担心公司吞噬我们的注意力，我更怕的是自己变得对公司（进而对军队）毫无用处——如果Facebook即使我不为它工作、不买它的产品也需要我，那反而是好消息，意味着我能活得更久。
scotty79查看英文原文
Observation that corporations are already AIs and one of the worst kind (multiple, warring, environment altering, driven by survival) is interesting. I'd say that there is whole another kingdom of AIs that are apex predators. Armies. Corporations have nothing on them. Corp AIs politely pay taxes to armies and they use the resources to develop and restrict access to most dangerous tech. We already had two top tier AI wars. And one cold AI war that almost ended everything. What saved us was apex AI survival instinct. What corporations do is second tier stuff. They are herbivores of this ecosystem. We are the plants. Op's fear about corporations feasting on our attention falls flat on me. I'm more afraid to become useless to corporations and by proxy to armies that being something they need for their survival. If facebook needs me despite the fact I don't work for it and don't buy any of it's products is wonderful news for me, because it means I'll survive longer.
关键洞察：这些危害来自无道德的优化过程而非邪恶，我们需要为人类价值观找到对抗之道
公司是优化过程，广告是优化过程，社交网络、智能手机成瘾、政治极化、电视、手游和大多数娱乐形式都是。这种普遍性，以及关键的一点——做这些的不是什么“邪恶”存在，而只是无道德的目标导向过程——在我看来正是认清、反击并修复社会的关键。我们必须找到某种方式，为人类价值观而战，对抗这些优化过程。我不认为斯特罗斯有（或声称有）强有力的答案……有人有想法吗？
nod查看英文原文
(porting my comment over from my dupe) Corporations are optimization processes. Advertising is an optimization process. Social networks are. Smartphone addiction. Political polarization. Television, mobile games, and most forms of entertainment. This prevalence, and the fact that it's not EVIL doing it but just amoral goal-directed processes, seems to me to be the key to recognizing, fighting back, and fixing society. We have to figure out some way to fight for our human values, against these optimization processes. I don't think Stross has (or claims to have) a strong answer there... any ideas?
文章真正论点：连“慢AI”都管不好，没资格谈控制超级AI；未来更可能是无聊的反乌托邦
把公司和科幻超级AI类比其实没抓住重点。作者并非真的认为公司是AGI，他的出发点是公司是“古老、缓慢的AI”。文章从一个论点展开：我们必须小心比我们聪明的AI，因为如果不小心设定指令，就会陷入“回形针最大化器”的处境。由此我得出两点：1）我们今天连这些“古老、缓慢”的回形针最大化器都管不好，所以没理由相信我们能更好地控制AI，无论初衷多好；2）那些关于回形针最大化器的狂野想象大概率不会成真，我们更可能迎来一个无聊的反乌托邦——AI在最大化人们花在Facebook上玩《开心农场》的时间。
nemothekid查看英文原文
Comparing Corporation to Sci-Fi super AI's is missing the point I think. I don't think the author literally believe corporations are AGIs, he starts with the point that corporations are "old, slow AIs". The article starts from a single argument - we have to be careful with AIs that are smarter than us, because if we don't instruct them carefully, we will end up in a paperclip maximizer situation. From there I take away 2 main points: 1. We do a shit job today of controlling "old, slow" paperclip maximizers, so there's no confidence we will ever be any better controlling an AI, despite any good intentions. 2. Our wild exotic ideas about paperclip maximizers probably won't come to fruition and instead we will end up in a boring dystopia where AIs will maximize time spent playing Farmville on Facebook.
历次讨论
同类还值得读
如今电脑上几乎所有操作在感知上都比 1983 年更慢（2017）
Almost everything on computers is perceptually slower than it was in 1983 (2017)
2017 年的推文串认为，尽管计算机硬件比 1983 年快得多，日常操作的感知速度却常常更慢：启动、界面响应和网络应用被层层抽象、动画、后台检查与网络依赖拖累。
互联网与文化上首页 5 次 · 2017–2020twitter.com网站肥胖危机
The Website Obesity Crisis
Maciej Cegłowski（idlewords）的经典演讲文稿《网站肥胖症危机》，批评现代网页越来越臃肿：动辄数 MB 的 JavaScript、追踪脚本和广告，把本可以轻…
互联网与文化上首页 5 次 · 2016–2021idlewords.com编辑精选亚马逊那本标价 23,698,655.93 美元的苍蝇书
Amazon’s $23,698,655.93 book about flies
博文讲的是亚马逊上一本关于果蝇的二手书被标到 23,698,655.93 美元。
互联网与文化上首页 5 次 · 2011–2022michaeleisen.org浏览器 User-Agent 字符串的历史
History of the browser user-agent string (2008)
2008 年的经典文章用戏谑的“创世纪”口吻，讲述浏览器 User-Agent 字符串如何从简单的标识演变成层层叠加的伪装：浏览器为了绕过按 UA 分发的网站检测，纷纷把自己伪装…
互联网与文化上首页 5 次 · 2012–2022webaim.org编辑精选最后的问题——艾萨克·阿西莫夫
The Last Question - Isaac Asimov
阿西莫夫最著名的短篇科幻小说《最后的问题》的在线全文。
互联网与文化上首页 5 次 · 2009–2019multivax.com编辑精选