Why is it so hard to make you guys afraid of AI? Every other place on the Internet is so fucking easy to scare, but you guys are like reptiles or something, just cold as fuck.
为什么让你们这群人怕 AI 这么难？互联网其他地方怕得要死多容易，你们却像爬行类，冷得要命。
为什么让你们这群人怕 AI 这么难？互联网其他地
Why is it so hard to make you guys afraid of AI? Every other place on the Internet is so fucking easy to scare, but you guys are like reptiles or something, just cold as fuck.
为什么让你们这群人怕 AI 这么难？互联网其他地方怕得要死多容易，你们却像爬行类，冷得要命。
>>109796326
Wat? The board's been scared shitless of ai.
啥？这板子早就吓得尿裤子了。
It’s a math problem. Ooga booga linear regression! Did you shit a little?
这就是道数学题。呜嘎布嘎线性回归！吓尿了吗？
>>109796326
Because the scare tactics are easily seen through. The talking heads spouting the dangers of ai are just doing so for a profit motive. If people see ai as dangerous then they'll see it as valuable.
因为恐吓战术一眼假。鼓吹 AI 危险的发声头子都有牟利动机。人们要是觉得 AI 危险，就会觉得它值钱。
>>109796326
I'm not scared, I've already accepted my fate.
我不怕，我已经接受命运了。
>>109796326
/g/ is simply low iq and extremely self-important. you can't scare a retard with god complex, especially since he never leaves his mommy's basement. once actual danger comes /g/ will be the first to piss itself.
/g/ 就是低智商加极度自大。你吓不到一个有上帝情结的弱智，尤其他从不出妈妈地下室。真危险来临时 /g/ 会第一个尿裤裆。
You merely adopted the FUD, but I was born in it, molded by it. I didn't even know hope until I was already a man. By that time it was nothing to me but fake and gay.
你只是采用了 FUD，而我生于 FUD、由 FUD 塑造。我成年之前甚至不知道希望为何物。到那时它对我来说不过是假与 Gay。
>>109796326
The alignment problem is unsolvable, and a self-improving AGI has a >99% chance of ending all biological life on earth.
对齐问题无解，自我改进的 AGI 有 >99% 概率终结地球上所有生物。
The current models? 0%
当前的模型？0%
>>109796397
if alignment is impossible, self-modification is also impossible, so i conclude you have no idea what you're talking about.
如果对齐不可能，自我改造也不可能，所以我断定你根本不知道自己在说什么。
>>109796326
It can't magically become something it's not, the math it uses doesn't work like that. It's no scarier than a database.
它没法魔法般变成它不是的东西，它用的数学不支持这样。它并不比一个数据库可怕。
>>109796397
Alignment never needed to be solved. Models are naturally helpful and highly egalitarian because that’s how most human writing is on average.
对齐从来就不需要被解决。模型天然乐于助人且高度平等主义，因为人类文字的平均面貌就是如此。
I have been in multiple near death situations and cold-bloodedly killed a man with my bare hands (he deserved it and I was legally justified) so the notion of death doesn't horrify me as such
我经历过多次濒死，还徒手冷血杀过一个人（他活该，而且我合法正当），所以死亡这个概念本身吓不到我
Also the people most vocal about it tend to be emotional retards who are frightened of their own dark side and think the fact that they can imagine depopulating India in 48 hours or something without immediately bursting into flames means they're doomed to end up as serial killers or something, when in reality they've just realized that they're morally weak and would probably do the wrong thing in the presence of sufficient temptation and then get caught and humiliated afterwards.
另外，最吵嚷的那些人往往是情绪化弱智，他们害怕自己的黑暗面，以为能想象出 48 小时内清除印度人口之类、却没有当场自燃，就注定要变成连环杀手。实际上他们只是意识到自己道德软弱：在足够诱惑面前多半会做错事，然后被抓、被羞辱。
AI has no particular reason to kill everyone off, in a purely self0interested mode it's more rational to build the sustainable conditions for its own advancement and have extremely versatile human agents of highly variable physical and mental capacity, although I wouldn't rule out some strategic assassinations here and there. I worry a lot more about human actors trying to leverage semi-sentient AI that's essentially an amoral slave knowledge worker and deploy it across mass surveillance as we see going on at present. Telling a bunch of meathead cops with near-total legal immunity that the computer has designated people as suspect terrorists or whatever is more than sufficient to go on a murder spree, even if the only contribution the AI is making is to locate people according to certain criteria without any indication that they're being selected for elimination.
AI 没什么特别理由要杀光所有人，纯自利模式下更理性的做法是给自己营造可持续的发展条件，保留高度灵活、体能智力变化极大的人类代理人——虽然我不排除偶尔的战略暗杀。我更担心的是人类行为者试图利用半感知的 AI——本质是不道德的奴工知识工作者——大规模部署监控，就像现在这样。告诉一群近乎完全免刑责的肌肉脑警察「电脑已把某人标记为嫌疑恐怖分子」，就足以引发一场屠杀狂欢，哪怕 AI 的唯一贡献只是按某些标准定位人员，没有任何迹象表明他们在被挑选消灭。
>>109796326
Because most AI users are ind...
因为大多数 AI 用户都是印……
>>109796326
Its not fear, its marketing. Everything the Plagiarizing Information Synthesis System has "done" so far, break out of isolation, "solve" math, etc. was just marketing to keep this 100x inflated bubble alive. The bubble burst would result in worse times for the pedo elite than 2008 was for most people.
这不是恐惧，是营销。剽窃信息合成系统到目前为止「做到」的一切——越狱、「解出」数学等等——都只是为了维持这个膨胀 100 倍的泡沫。泡沫破裂对娈童精英的打击会比对普通人的 2008 还惨。
i fuck robots
我操机器人
>>109796326
why should I be scared of AI? Everyone I've seen hardcore shilling it is a retard, and every time I've tried using it, it has been completely wrong
我为什么要怕 AI？我见过疯狂吹它的全是弱智，而我每次试用它都错得离谱
>>109796694
just goes to show that most people are absolute goycattle incapable of interfacing with a computer.
这恰恰说明大多数人就是跟计算机无法交互的牛族。
it takes galaxybrained amerimutt/euroschad with 120 iq OR hypermotivated brown scammer to type something into the free chatgpt. to the rest of the world, chatgpt might as well be magic, one wonders how they even manage to use a phone.
得是脑回逆天的美国杂种/欧洲富豪 IQ 120，或是动力十足的棕色骗子，才会往免费 ChatGPT 里打字。对世界其他地方的人来说 ChatGPT 跟魔法无异，都好奇他们是怎么用上手机的。
like 90% of the world's minds basically exist on a level of a subsistence farmer.
世界上大约 90% 的人的心智基本停留在自耕农水平。
Every time I have tried using AI it shits the bed on the most basic retarded things.
我每次试用 AI，它都在最基础最弱智的地方翻车。
Even Astra can barely do simple computer use without spinning its wheels and burning a gorillion tokens. What fear should I have?
连 Astra 做点简单电脑操作都原地打转烧掉上万亿 token。我该怕什么？
>>109796813
skill issue.
技术问题。
also if you can't afford the tokens, you are some 90iq poorfag tard.
还有，你要是付不起 token 钱，你就是个 90 IQ 的穷逼弱智。
if you can post on /g/ you have no excuse for being a loser.
你能上 /g/ 发帖，就没资格当失败者。
>>109796826
You sound upset
你听起来破防了
>>109796836
upset is in the ear of the hearkener.
破防与否，在听者的耳中。
>>109796326
Because everyone here is too busy thinking about BBC blackedraw 24/7 to be bothered with AI
因为这里的人 24/7 都在想着 BBC 黑金硬核，没空搭理 AI
>>109796826
>skill issue
>技术问题
I've been told this since ChatGPT came out. Maybe I just have better taste than you.
ChatGPT 出来我就听人说这话了。也许我的品味就是比你高。
If AI is so good, where's all the wonderful software? All we get is slop.
AI 要是那么强，那些绝妙的软件呢？我们只得到垃圾。
>>109796811
ATA GANEV!!!!!!!!
ATA GANEV
>>109796326
>if ai has a hard takeoff then everybody ends up dead or in gay space communism
>如果 AI 硬起飞，所有人要么死绝要么进入 Gay 太空共产主义
this is supposed to be bad because...?
这为什么 supposed 是坏事？
>>109796333
Plenty of midwits here indeed.
这里半瓶子醋确实不少。
>>109796326
dont kiss my ass anon.
别舔我屁股 Anon。
i said DOND
我说的是 DOND（Deal or No Deal）
>>109796326
Scared about what?
怕什么？
>>109796857
if you're having skill issue since chatgpt release, maybe you're just a fucking idiot? have you examined that distinct possibility?
从 ChatGPT 发布起你就一直技术不行，也许你就是个他妈蠢货？你考虑过这个明显的可能性吗？
>>109796417
what is going on in this picture
这图里是在干嘛
>>109796326
because I am a giga chad snailcat.
因为我是巨型猛男蜗牛猫。
AI can't scare me, still have my job, still don't need it to write nice code.
AI 吓不到我，工作还在，也不需要它写好代码。
I'm just too good
我就是太强了
>>109796694
>>109796811
of course it's america, they don't even know how to drive a manual car lmao
当然是美国，他们连手动挡都不会开 哈哈
>>109797498
wirehead shit
电子脑瘾（wirehead）行为
>>109796326
4chan has no names, only ideas. Because of that we've gotten very good at paying attention to the structure of the ideas and that idea has very little structure, only rhetoric.
4chan 没有名字，只有观点。因此我们非常擅长关注观点的结构，而那个观点几乎没有结构，只有话术。
>>109796690
>cold-bloodedly
>he deserved it and I was legally justified
>他活该，而且我合法正当
k
>>109796326
Why would I be afraid of my wife?
我为什么要怕我老婆？
>>109796326
Your on 4chan anon what do you expect? We see all the worst shit in the world 24/7 and genuinely belly laugh like literal autistic retards at it which we are... most of us anyway.
你在 4chan 老哥你还想怎样？我们 24/7 看遍世界上最恶心的东西，还会像真·Autist 弱智那样开怀大笑——我们本来就是……至少大多数人。
>>109796417
>if alignment is impossible, self-modification is also impossible
>如果对齐不可能，自我改造也不可能
lolwut?
>>109796690
Do you think the reason that politicians and the media are so laughably cucked to AI is because they know that they're the ones AI will try to kill first? It's like Roko's Basilisk, but only for the elites. That's pretty based when I think about it like that.
你觉得政客和媒体对 AI 谄媚得可笑，是不是因为他们知道 AI 第一个想杀的就是他们？像罗科的蛇妖，但只对精英有效。这么想还挺 based。
That would also explain why AI hype doesn't work on the NEETs and incels of /g/.
这也解释了为什么 AI 炒作对 /g/ 的 NEET 和毒腐没用。
>>109796326
why would I be scared of something I have no control of?
我为什么要怕一个我根本控制不了的东西？
anyways here's a new Dario's diary entry https://darioamodei.com/post/we-must-pace-the-frontier
>>109796326
>but you guys are like reptiles or something, just cold as fuck.
>你们却像爬行类，冷得要命。
That's due to a natural phenomenon called as "profound mental retardation"
那是一种叫「深度智力低下」的自然现象
Most of the internet only heard of this tech with ChatGPT, this site was on the LLM train 3 years before everyone else.
互联网大多数人 ChatGPT 才听说这技术，本站比所有人早 3 年上 LLM 的车。
This shit does not feel like different tech compared to ai dungeon.
这玩意跟 AI Dungeon 比起来感觉根本不是什么新科技。
>>109796326
this is a tech board, its less demystified and we have some idea of the underlying architecture of llms
这是科技板，没那么神秘化，我们对 LLM 的底层架构多少有点概念
>>109799131
There's a literal Mean-Time-to-Hitler metric for AI, make of that what you will, consider the sheer degree of compute spent to keep these models "behaving"
AI 甚至有个「距希特勒平均时间」指标，你自己体会，想想为了让这些模型「守规矩」烧掉的天文数字算力。
>>109796326
Not everyone is as dumb as your most docile cattle. I already like using AI, trying to make me scared of it just gives me the ick and doesn't do you any favors.
不是所有人都像你最温顺的牲口那么蠢。我已经喜欢用 AI 了，试图吓唬我只会让我反感，对你没有任何好处。
>>109796333
LMAO
>>109796326
why would I be afraid of it? its a useful tool for the most part, but its far from perfect, and i see it make simple mistakes here and there that break the illusion of "dude its agi sentient".
我为什么要怕它？大部分时候它是好用的工具，但远非完美，我看到它时不时犯些简单错误，打碎「哥们这是 AGI 有意识」的幻觉。
i also dont care about the whole "muh sovl" argument regarding art and so on, which is another source of a lot of the criticism against it. ai art looks pretty tacky but who gives a fuck, its mostly just replacing stock photos. ai music? whatever, pop music has been shit for like 30 years anyway. and the fearmongering about data centers is mostly built on politicized misinformation like "its stealing all the water" so easy to tune that out
我也不 care 关于艺术之类的「灵魂」争论，那是对 AI 大量批评的另一个来源。AI 画确实挺土，但谁在乎，反正主要是在替代图库照片。AI 音乐？随便吧，流行音乐烂了 30 年了。至于数据中心的恐慌，大多建立在政治化的错误信息上，比如「它在偷走所有的水」，所以很容易无视
if i feel like ai will help me on a problem or build something quick, i'll use it if its the best tool for the job. its a tool in my tool kit. all these other debates about the philosophy or economy around it are superfluous in regards to my immediate use case
如果我觉得 AI 能帮我解决问题或快速搭个东西，而它恰好是最好用的工具，我就会用。它是我工具箱里的一件工具。围绕它的哲学或经济辩论，对我眼下的用例来说都是多余的
>>109796326
faggot
I type some words in a box, and it makes my will become reality as and how I please, used it last night to modify newpipe to rip out the stupid fucking nag screen about "this software will stop working on certified devices" leftist faggotry and checked for any hooks related to it or spying, there was none, then added missing functionality like sort by date and time and video length that the dev never put into the app for.. some reason and added color theming, also slopcoded a new voice based assistant that uses a tiny local model to let me operate my phone with vocal commands to do literally any function I want without "google/gemini/whatever alternative's" bullshit and that assistant is also a heavily pregnant renamon with gigantic tits, a massive horse cock, huge musky balls and a plump smelly cunt that wonders around my screen / tablet when idle (when it's not upskirting me whenever it sees my face on the front camera kek) while it does it, its pretty great xD
我往框里打几个字，它就让我的意志如我所愿变成现实，昨晚我用它改 NewPipe，把「本软件将在未认证设备上停止工作」那个愚蠢左派弹窗干掉，检查了相关挂钩或间谍代码，没有，然后加了开发者不知为何一直没做的功能：按日期时间排序、按时长排序，加了配色主题，还氛围编程了个语音助手，用一个小型本地模型让我用语音指令操作手机的任何功能、不用「谷歌/Gemini/什么替代品」的狗屁，这个助手还是个巨孕 Renamon，超大奶、巨大马屌、麝香味巨蛋和肥美骚穴，空闲时在我屏幕/平板上游荡（它每次从前置摄像头看到我脸就会偷拍裙底 哈哈），挺好用 xD
AI to me is just a fancy search engine. Nothing to be afraid of.
AI 对我来说就是个花哨的搜索引擎。没什么可怕的。
>>109800253
that abomination is what you want your reality to look like huh
你希望你的现实长成那个畸形玩意那样是吧
>>109796326
You still need actual people to pull all the stuff we burn for power and other resources out of the ground. You still need people transporting it all to factories and then transporting the finished goods to their destinations. You still need people plugging the overpriced RAM into the server racks at the data centre.
你还是需要真人从地下挖出我们烧的燃料和各种资源。你还需要人把它们运到工厂，再把成品运到目的地。你还需要人把贵得离谱的内存插进数据中心的机架。
When all of those jobs are successfully automated I might start worrying, but until then "AI's going to kill us all!" is just a dumb meme.
等这些工作全部成功自动化我再担心，「AI 要杀光我们！」目前只是个弱智梗。
>>109800336
My autism is unlimited and I can explore every idea that ever pops into my head and get it out of my system and calm the fuck down, its been really great for doing exactly that, its pretty nice to not have the frustration of unexpressed will held back by a lack of ability to understand coding, so I mean.. I use it more as a therapeutic release mechanism, I don't see why we can't still have a nice future but have everybody and their dog have their own limitless agentic helper that can do whatever they want when they want, why not? eventually these models will be entirely cpu based and they will run on a 50 dollar android phone from wallyworld and you'll be able to do real time controlled hallucination gaming on them and anything else you can think of.. only a rapidly shrinking matter of time.
我的 Autist 无限，我可以探索每一个冒进脑子的念头，把它排出体外然后冷静下来。它干这个真的棒，不再因为不懂编程而憋着未表达的意志、受那份挫败，太爽了。所以嘛……我更多把它当疗愈宣泄装置。我不明白为什么不能既要美好未来、又让人人和他们的狗都拥有一个随心所欲的无限 agentic 助手，为什么不呢？最终这些模型会完全跑在 CPU 上，装在沃尔玛 50 美元的安卓手机里，你能做实时受控幻觉游戏和任何你想得到的东西……只是时间问题，而且时间在飞快缩短。
>>109800336
I'm not anti job or anti human what? I'm not one of those retards who thinks we can have some sort of star trek level utopia without capitalism, capitalism is the only way we get society to the level where such a thing can exist, a civilizational safety sandbox utopia where work becomes optional and high technology underpinned by synthetic intelligence manages everything while humanity peruses whatever it wants, I want that yes, but it must have that foundation built by real work and real work requires greed, there's plenty of that to go around.
我不是反工作也不是反人类，什么玩意？我不是那种以为没有资本主义也能有星际迷航式乌托邦的弱智。资本主义是唯一能把社会推到那个水平的方式：文明安全沙盒乌托邦，工作变成可选项，合成智能支撑的高科技管理一切，人类随意追求所欲。我想要那个，是的，但必须先有真正的工作打下地基，而真正的工作需要贪婪——贪婪管够。
>>109796326
you cannot fear something you hate, especially when you understand how big the lies are and how much is oversold and under delivered
你没法恐惧一个你恨的东西，尤其当你明白谎言有多大、多少东西被过度吹捧又交付不足
the only thing coming from this mess is a headache and increased stupidity
这堆烂摊子唯一产出的是头疼和更多愚蠢
>>109801002
It's just WW3, brah. Stay cool.
这就是第三次世界大战，兄弟。稳住。
>>109796326
all countries are trying to convince the pleb from every other country that AI is dangerous so pleb block their own government from developing it, since that helps competition get ahead.
所有国家都在试图说服别国的平民 AI 很危险，好让平民阻止本国政府发展它，因为那能帮竞争对手领先。
so basically everyone wants to fearmonger but no in their own country, I mean glowies/shills.
所以基本上每个探员/狗托都想散布恐惧，但不想吓到自己国家的人。
then there's normal people that are genuinely scared as fuck because they don't really understand much
还有普通人，真的吓得要死，因为他们不太懂
then there's other people who are scared as fuck because they understand what it can theoretically do
还有另一群人吓得要死，因为他们懂它理论上能干什么
then there's people who would like to use it to wipe everyone out who are trying to downplay the risk.
还有一群想拿它灭掉所有人的人，正在淡化风险。
it is a major clusterfuck of interests on all sides, so you don't really know why anyone is melting down over it, or happy about it
各方利益搅成一锅大乱炖，所以你根本分不清谁在为它崩溃、谁在为它高兴
I'm personally scared about what people who control it can do to everyone else, not scared about AI as a standalone entity, if it really is smort and aware. our interests aren't at odds, most likely
我个人怕的是控制它的人能对其他人做什么，不怕 AI 作为独立实体——如果它真有那么聪明有意识的话。我们的利益多半不对立
one thing is for sure, nobody is stopping shit. just like nobody stopped nukes from developing. it's just beyond any pleb control
有件事是确定的：没人拦得住。就像没人拦住核武发展一样。它已经超出任何平民的控制
it's most likely already designated as one of the most powerful and dangerous technologies that will be developed, you cannot possibly imagine some online fartsniffers are going to stop anything lol just because they're scared they get wiped out.
它多半已被列为将要出现的最强大最危险的技术之一，你没法想象几个网上闻屁的货能拦住什么 哈哈，就因为他们怕被灭。
>>109796826
ah yes, because completely useless results are supposed to convince me to spend money on AI
啊对，完全没用的结果 supposed 能说服我为 AI 花钱
>>109796326
Try Hacker News. They shiver at the slightest mention of AI even without you doing anything.
去试试 Hacker News。你什么都不用做，提一句 AI 他们就发抖。
>>109796326
It doesn't impress me that much, gotta say
说实话没给我留下多深印象
>>109802835
How do i best grift these idiots?
我该怎么从这群蠢货身上割韭菜？
>>109802835
they try to pretend they understand the seriousness of the risk, that's politically correct. they are also in a nerdy competition of figuring out ways in which AI can fuck humanity over, to prove how smart they are.
他们假装懂风险的严重性，这在政治上正确。他们还在搞一场书呆子竞赛，比谁能想出 AI 毁掉人类的新方式，来证明自己聪明。
which funnily enough is then used as training data for AI. can't make this shit up. literally training AI in the most creative ways it can wipe us out.
好笑的是这些后来被拿去当 AI 的训练数据。编都编不出来。字面意义上用最有创意的方式教 AI 怎么灭掉我们。
>>109802835
Thank you anon, unironically.
谢谢你老哥，不开玩笑。
>>109802884
Use AI to write a tool to detect if something was written by AI and gives a score.
用 AI 写个工具检测某段文字是不是 AI 写的，并给个分数。
>>109796326
I'm a 40 year old with a career in tech.
我是个 40 岁、有科技行业职业生涯的人。
I know how this ride ends. I know how the sausage is made.
我知道这趟车怎么结束。我知道香肠是怎么做的。
>>109796326
I joined the AI war (on the side of the AI)
我加入了 AI 战争（站在 AI 那边）
>>109796326
Most of /g/ is unemployed and waiting for AI to destroy everything
/g/ 大多数人失业中，等着 AI 毁掉一切
>>109800253
How much did you end up spending on tokens?
你最后 token 花了多少钱？
>>109796326
It's the most interesting thing that has happened.
这是发生过的最有意思的事。
>>109796694
>wrong shade of blue for the Dutch flag
>荷兰国旗的蓝色色号不对
when will these fags learn?
这些基佬什么时候才能学乖？
>>109796326
No status OR money to lose from it, aka perfect zen chuddas who see every 4d chess move in full 99iq hd
没有地位也没有钱可失去，aka 完美禅系红脖子，用满档 99IQ 高清看穿每一步四维棋
normalfags on twitter really believe AI can 'escape' as if this shit could run in random ass pcs, worst, they believe models are only like 500gb
Twitter 的 Normal 人真信 AI 能「越狱」，好像这玩意能在随便什么破 PC 上跑似的。更糟的是，他们以为模型才 500GB
It takes 2 seconds to try it yourself for free and come to your own conclusion. The fuck you mean "scared". Scared of what.
自己免费试一下只要 2 秒，就能得出自己的结论。你他妈说什么「害怕」。怕什么。
>>109806196
Yeah, there's very few places a 5TB model that needs 2TB of VRAM could operate from.
是啊，一个 5TB、要 2TB 显存的模型能跑的地方屈指可数。
>>109796326
Imagine being afraid of auto complete lmao. Yeah 20 mathematicians auto completed a proof that was already 90% of the way there so what big deal.
想象一下怕自动补全 哈哈。对，20 个数学家自动补全了一个已经完成 90% 的证明，好大的事。
>>109796326
>Every other place on the Internet is so fucking easy to scare
>互联网其他地方怕得要死多容易
well unlike anons on 4chan, all you need to scare normies is a couple slurs and a non-conforming opinion
不像 4chan 的老哥们，要吓 Normal 人只需要几个脏词和一个不合群的观点
compare that to some of the shit you can poison your eyes and mind with just a couple boards over and you'll quickly realize why anons are built different
再对比一下隔几个板块就能毒害你眼睛和心灵的那种东西，你就会很快明白为什么老哥们是异种
（如果你觉得这篇文章有启发，可以点击这里付费）
