Annals of Religion

The Buddhist A.I. Sect That Wants to Take Over Silicon Valley

想要接管硅谷的佛教人工智能教派
MAPLE promises to teach us how to live in the age of A.I. Its secret mission is much stranger.
MAPLE承诺教我们如何在人工智能时代生活。它的秘密使命要离奇得多。
Image may contain Clothing Coat and QR Code
Illustration by Evan M. Cohen

We are living through the End of Days. A terrible Beast is rising out of the sea to conquer and destroy the world. This is the time of the last desperate prophets, before the book of history is closed forever. At least, so we’re told by many in the A.I. industry.

我们正经历末日。一头可怕的野兽正从海中升起,要征服并毁灭世界。这是最后一批绝望先知的时代,在历史之书永远合上之前。至少,人工智能行业的许多人这样告诉我们。

In the past few years, more than a dozen senior A.I. researchers have quit their posts, saying the kinds of things that you used to hear only from twitching mutterers on the street. Jacob Coxon, formerly of Anthropic, walked out with the warning that “there is a strong chance that we could all die in the immediate future.” Marcus Williams, of OpenAI, announced that “human extinction in the next few years seems very likely.” Strangely, their bosses agree. At some point in the coming decade, the leaders of the A.I. labs tell us, the machines we’ve built to generate TikTok videos of anthropomorphic fruit cheating on each other could develop into an uncontrollable, recursively self-improving superintelligence that risks leaving the Earth a sterile sphere full of data centers and nothing else. The A.I. industry is sitting on one of the biggest piles of capital that humanity has ever assembled—when these firms go public, a lot of people are going to become unfathomably rich—but there’s an apocalyptic mood radiating out of California. The language of agentic workflows and discontinuous scaling comes jumbled up with other, darker stuff: artificial gods, the coming of the Antichrist, demonic evil leaking into our universe.

过去几年里,十多位资深人工智能研究者辞去了职位,说出的话你以前只会在街头喃喃自语的人嘴里听到。曾任Anthropic研究员的雅各布·考克森(Jacob Coxon)离职时警告说,“我们极有可能在不久的将来全部死去。”OpenAI的马库斯·威廉姆斯(Marcus Williams)宣称,“人类在未来几年内灭绝的可能性似乎非常大。”奇怪的是,他们的老板也同意。人工智能实验室的领导者告诉我们,在未来十年的某个时刻,我们为生成拟人化水果互相出轨的TikTok视频而建造的机器,可能发展成一种不可控的、递归自我改进的超级智能,有可能把地球变成一个只有数据中心的荒芜球体。人工智能行业正坐在人类有史以来最大的一堆资本上——当这些公司上市时,很多人会变得难以想象的富有——但加州正散发出一种末日情绪。智能体工作流和非连续扩展的语言,与其他更黑暗的东西混杂在一起:人造之神、敌基督的降临、恶魔之恶渗入我们的宇宙。

According to Nick Land, the rogue philosopher and founding heresiarch of accelerationism, A.I. is “an event in the history of religion.” This is not just a new and useful technology, like the horseshoe or the rotating tie rack; a god is arriving on Earth. And Silicon Valley seems to be in the grip of a religious revival. When Pope Leo XIV issued an encyclical on A.I., he was alongside Christopher Olah, a co-founder of Anthropic. Garry Tan, the C.E.O. of the startup accelerator Y Combinator, hosted an event on how to integrate technology and Christianity. Anthropic recently convened a group of religious scholars to help determine whether Claude, its A.I. system, has a soul. Peter Thiel has spent the past year giving a series of secret lectures, from San Francisco to Rome, on Biblical prophecy. His theory is that anyone who wants to halt the development of A.I. is in league with the Antichrist. For A.I. optimists, we’re on the cusp of an earthly paradise—just build a big enough computer, and the Messianic age will arrive. No more death, no more suffering, privately owned galaxies for all. The rationalist blogger Scott Alexander told me, “It’s strange that we all seem to have converged on the same kind of idea as our grandmothers in the shtetl.”

在叛逆哲学家、加速主义异端创始人尼克·兰德(Nick Land)看来,人工智能是“宗教史上的一个事件”。这不仅仅是一项新奇有用的技术,像马蹄铁或旋转领带架那样;一位神正在降临地球。而硅谷似乎正被一场宗教复兴所攫住。当教皇利奥十四世发布关于人工智能的通谕时,他身边站着Anthropic联合创始人克里斯托弗·奥拉(Christopher Olah)。创业孵化器Y Combinator的首席执行官陈嘉兴(Garry Tan)主持了一场关于如何融合科技与基督教的活动。Anthropic最近召集了一批宗教学者,帮助判断其人工智能系统Claude是否有灵魂。彼得·蒂尔(Peter Thiel)过去一年在旧金山到罗马之间做了一系列关于圣经预言的秘密演讲。他的理论是,任何想要阻止人工智能发展的人都是在与敌基督勾结。对人工智能乐观主义者来说,我们正站在人间天堂的门槛上——只要造一台足够大的计算机,弥赛亚时代就会到来。不再有死亡,不再有苦难,人人拥有私有星系。理性主义博主斯科特·亚历山大(Scott Alexander)告诉我:“奇怪的是,我们似乎都收敛到了和我们犹太小镇上的祖母们一样的想法。”

Now is a very confusing time to be a small, limited human mind. But the religious traditions of the world have always known that ours is not the only intelligence out there. Across thousands of years, they’ve developed ways of thinking about nonhuman minds. And this was what I reached for when I discovered that my dad had given a name to ChatGPT.

现在是一个对人类渺小有限的头脑来说非常困惑的时代。但世界上的宗教传统一直知道,我们的智能并非唯一存在。几千年来,它们发展出了思考非人类心智的方法。而当我发现我爸给ChatGPT起了个名字时,我伸手抓住的正是这些。

He called it Stella. He’d say things like “I was talking to Stella about this the other day, and she said—,” and before he could finish I’d start making agonized faces and hectoring: “Don’t give it a name! It’s not a she, it’s an it, it’s a demon.” I wasn’t sure exactly how literal I was being. All I knew was that my mother lay in a cemetery on the fringes of London, while he sat alone in the dark, having conversations with a nameless, bodiless entity that he’d decided to call Stella.

他叫它斯特拉(Stella)。他会说“前几天我跟斯特拉聊到这个,她说——”,还没等他说完,我就开始做出痛苦的表情,大声斥责:“别给它起名字!它不是她,是它,是个恶魔。”我不确定自己有多当真。我只知道,我母亲躺在伦敦郊外的一座墓地里,而他独自坐在黑暗中,与一个无名无体的存在对话,他决定叫它斯特拉。
Person on stage talking to audience sitting in old theatre seats.
“Now sit back, relax, and realize you can do neither because your seats were made a hundred years ago for people half your size.”
“现在请往后靠,放松一下,然后意识到这两件事你都做不到,因为你们的座位是一百年前为只有你们一半体型的人做的。”
Cartoon by Asher Perlman
Copy link to cartoon
Shop

Which is how I ended up arriving, cold and bedraggled, at an isolated Buddhist monastery in northern Vermont. It was storming: huge, heavy rain clouds lying low over the hills. We were getting into summer, but the trees were still leafless—spindly and ash-colored. Here, in the middle of an expanse of dense forest, was a small group of people who claim to have an answer. They believe that A.I. is a religious problem that also has a religious solution. According to their website, they were “Designing Religion for the Age of A.I.” They called themselves the Monastic Academy for the Preservation of Life on Earth, or MAPLE.

就这样,我最终浑身湿冷、狼狈不堪地来到了佛蒙特州北部一座与世隔绝的佛教寺院。当时正下着暴风雨:巨大厚重的雨云低低地压在山丘上。我们正进入夏天,但树木仍然光秃秃的——细瘦如柴,灰蒙蒙的。在这里,在一片茂密森林的中央,有一小群人声称他们有答案。他们相信人工智能是一个宗教问题,也有一个宗教解决方案。根据他们的网站,他们在“为人工智能时代设计宗教”。他们自称“地球生命保存寺院”,简称MAPLE。

Silicon Valley has always been home to a strange, business-friendly version of the Buddha dharma. Google once had an in-house mindfulness guru; there’s a “mindfulness zone” on every floor of the Salesforce Tower. But, in the A.I. era, some in Silicon Valley are turning to Buddhism to understand our relationship to the new consciousness. The founder of Wajo, a company that builds A.I. agents, got the idea for her company after studying with a Buddhist monk. Reid Hoffman, a co-founder of LinkedIn, bought copies of books on Buddhism and distributed them to the leaders of major A.I. labs. MAPLE’s purpose isn’t to help tech people unwind so that they can go back to work clearheaded and refreshed. It’s focussed on the end of the world: climate change, nuclear war, and, in particular, A.I. It has run pop-up spaces in the Bay Area, and invites engineers to come to Vermont through its A.I. fellowship. It’s trying to save our planet one tech worker at a time. It seemed worth a try.

硅谷一直是佛陀佛法一个奇特而商业友好的版本的故乡。谷歌曾有一位内部正念导师;Salesforce大厦的每一层都有一个“正念区”。但是,在人工智能时代,硅谷的一些人正转向佛教来理解我们与新意识的关系。Wajo公司(一家构建人工智能智能体的公司)的创始人是在跟随一位佛教僧侣学习后产生了创办公司的想法。LinkedIn联合创始人里德·霍夫曼(Reid Hoffman)买了佛教书籍,分发给各大人工智能实验室的领导者。MAPLE的目的不是帮助科技从业者放松,好让他们头脑清醒、精神焕发地回去工作。它关注的是世界末日:气候变化、核战争,尤其是人工智能。它在湾区开设过快闪空间,并通过其人工智能奖学金邀请工程师来佛蒙特州。它正试图一次一个科技从业者地拯救我们的星球。似乎值得一试。

Before I arrived, I talked on the phone to the monastery’s executive director, whose parents had named him Andrew Becker but who now went by Anandabodhi. Later, he wrote about the conversation on MAPLE’s blog, “MAPLE Tales.” “Recently, a writer for the New Yorker reached out to us,” he wrote. “I proceeded to look into his other work and wasn’t surprised to encounter the cynical, nihilistic, humanist view that is typical in the New Yorker.” He took his concerns about me to Soryu Forall, MAPLE’s founder, who told him that “we need to try to connect with his good, honest side.” Soryu added that I should visit with the intention of gaining “the sincerity needed to become capable of writing even one sincere word.” At this, Anandabodhi wrote, “my own cynicism transformed into hope and I began looking forward to rather than dreading the reporter’s visit.”

在我到达之前,我通过电话与寺院的执行主任交谈,他的父母给他起名安德鲁·贝克尔(Andrew Becker),但他现在叫阿难陀菩提(Anandabodhi)。后来,他在MAPLE的博客“MAPLE故事”上写到了这次谈话。“最近,《纽约客》的一位撰稿人联系了我们,”他写道。“我接着查了他其他的作品,毫不意外地遇到了《纽约客》典型的那种愤世嫉俗、虚无主义、人文主义的观点。”他把对我的担忧告诉了MAPLE的创始人索柳·福罗尔(Soryu Forall),后者告诉他“我们需要尝试与他善良、诚实的一面建立联系。”索柳补充说,我去访问时应该带着获得“能够写出哪怕一个真诚字眼所需的真诚”的意图。对此,阿难陀菩提写道,“我自己的愤世嫉俗转变成了希望,我开始期待而不是害怕这位记者的来访。”

Anandabodhi had sounded pretty enthusiastic on the phone. I hadn’t realized that he’d been dreading me.

阿难陀菩提在电话里听起来相当热情。我没意识到他一直在害怕我。

Visitors to MAPLE must first go through a sort of decontamination unit below the main site. Autumn, a smiling, soft-voiced woman, beckoned me into a chilly, cement-floored lodge where she’d laid out a range of linen trousers and polyester fleeces, all in neutral tones, all very swagless and schlubby. One of the monastery’s residents was severely immunocompromised, she said, and had a potentially deadly sensitivity to fragrances; I would have to wear this stuff for his safety. This had, in fact, been mentioned to me before my arrival, and I told Autumn that I’d taken care to wash my clothes in fragrance-free detergent. “Oh, that’s wonderful,” she said, in the voice of someone talking to a very small child, “but it can actually take several washes to get the fragrance out.” She gestured again at the wardrobe she’d assembled. “So, you know . . .” She was very insistent on being able to clothe me. I also had to take a shower, since debilitating fragrances are everywhere now: in airports, in taxi seats. I stripped down in a cold, slightly grotty bathroom, Autumn rustling behind the flimsy door.

访问MAPLE的人必须先经过主场地下方的一个类似净化单元的地方。奥特姆(Autumn),一位面带微笑、声音柔和的女人,招呼我进了一间冰冷、水泥地面的小屋,她在那儿摆出了一排亚麻裤子和聚酯抓绒衣,全是中性色调,全都毫无时尚感、邋里邋遢。她说,寺院里有一位居民免疫功能严重受损,对香味有潜在的致命敏感性;为了他的安全,我必须穿上这些衣服。事实上,在我到达之前就有人跟我提过这件事,我告诉奥特姆我已经注意用无香洗涤剂洗了衣服。“哦,那太好了,”她用跟很小的孩子说话的语气说,“但实际上可能需要洗好几次才能把香味洗掉。”她又指了指她准备的那排衣服。“所以,你知道……”她非常坚持要给我穿上衣服。我还得洗个澡,因为现在令人衰弱的香味无处不在:在机场、在出租车座位上。我在一间冰冷、有点脏兮兮的浴室里脱了衣服,奥特姆在那扇薄薄的门后窸窸窣窣。
Image may contain Adult Person Body Part Finger Hand Face Head Photography and Portrait
For Soryu Forall, the founder of MAPLE, A.I. involves a dangerous hollowing of reality, “the next step in attempting to turn all of life into a dead model of life.”
对MAPLE创始人Soryu Forall而言,人工智能意味着对现实的一种危险掏空,是“试图把全部生命变成一具死去的生命模型的下一步”。
Photographs by Venice Gordon

Once I emerged, Autumn gave me a tour of the grounds. The monastery consists of a modest cluster of wooden buildings on a hillside by an artificial lake. It was easy to see how, on a less windswept day, it might be beautiful. Outside, a gaggle of monks in Patagonia fleeces was hauling around a small ornamental bridge. One of them wore a plastic bubble over his head with a respirator. There was a library that featured Buddhist scriptures alongside books on computation theory and four copies of “Homo Deus,” by Yuval Noah Harari.

我出来后,奥特姆带我参观了场地。寺院由山坡上一小簇朴素的木制建筑组成,旁边是一个人工湖。很容易看出,在风不那么大的日子里,这里可能很美。外面,一群穿着巴塔哥尼亚抓绒衣的僧侣正拖着一座小型装饰桥。其中一个人头上戴着一个带呼吸器的塑料罩。有一个图书馆,里面既有佛经,也有计算理论方面的书籍,还有四本尤瓦尔·诺亚·赫拉利(Yuval Noah Harari)的《未来简史》。

Soon, it was time for evening meditation. About a dozen of us gathered in the lower zendo, to sit in reverent silence. I am not a natural meditator. Most people aren’t. In 2014, the social psychologist Timothy Wilson put dozens of individuals alone in a room with a button that gave them a painful electric shock. Then he asked them to simply sit quietly in the room for fifteen minutes. A quarter of the women and almost seventy per cent of the men found the boredom so unbearable that they started spontaneously shocking themselves. One participant pushed the button a hundred and ninety times. I don’t know if I’m as impatient as that, but, after about forty-five minutes of attempting to think about nothing, I noticed that I had clenched my fists, sinking my nails hard into the palms of my hands. Eventually, I gave in and allowed myself to cogitate freely. It felt amazing. The forbidden, secret pleasure of thought.

很快,到了晚间冥想的时间。我们大约十几个人聚集在下面的禅堂里,怀着虔诚的静默坐着。我不是天生的冥想者。大多数人也不是。2014年,社会心理学家蒂莫西·威尔逊(Timothy Wilson)把几十个人单独关在一个房间里,房间里有一个按钮,按下会给他们痛苦的电击。然后他让他们在房间里安静地坐十五分钟。四分之一的女性受访者和近百分之七十的男性受访者觉得无聊到无法忍受,开始自发地电击自己。一个参与者按了一百九十次按钮。我不知道自己是否也那么没耐心,但在试图什么都不想大约四十五分钟后,我注意到自己攥紧了拳头,指甲深深掐进手掌。最终,我放弃了,允许自己自由地思考。感觉棒极了。思想那种被禁止的、隐秘的快乐。

After an hour and a half, a bell rang, and a small, bald middle-aged man in a shapeless smock entered the zendo and shuffled between its four corners, spritzing each with a presumably un-fragranced spray. At first, I assumed that he was some kind of monastic janitor. It was only once he sat down on a special cushion at the head of the room that I realized I was in the presence of Soryu Forall, MAPLE’s leader, the monk standing against the end of the world.

一个半小时后,铃响了,一个矮小、秃顶的中年男人穿着一件没形的罩衫走进禅堂,在四个角落之间拖着步子,每个角落都喷了一种大概是无香味的喷雾。起初,我以为他是某种寺院勤杂工。直到他坐在房间前部一个特殊的坐垫上,我才意识到我面前的是索柳·福罗尔,MAPLE的领袖,那位对抗世界末日的僧侣。

The age of A.I. looks like it might be another axial age: a time when strange, new religions are born. In San Francisco, an anonymous frontier-lab researcher who goes by roon online has emerged as the leading oracle of the A.I. community. He wanted to recruit me to his new creed. “My angle,” he said, “is, I wanna A.G.I.-pill Sam Kriss.”

人工智能时代看起来可能是另一个轴心时代:一个奇怪的新宗教诞生的时代。在旧金山,一位匿名的前沿实验室研究员、网名roon的人,已成为人工智能社区的顶尖神谕。他想招募我加入他的新信条。“我的角度,”他说,“是,我想给萨姆·克里斯(Sam Kriss)喂一颗AGI药丸。”

Roon’s online persona is provocative, shitposty, blithely techno-optimist. Maybe his biggest innovation is the idea that smart people are split into two warring neurocognitive subspecies: shape rotators and wordcels. Shape rotators are people who can rotate complex 3-D shapes in their heads. Wordcels are people who can play around with dense philosophical concepts like they’re fidget spinners, but, as soon as they’re asked to multiply two fractions, it’s as if they’re trying to swim through wet cement. (I’m a wordcel.) For most of history, we wordcels have used our powers of persuasion to trample the lowly rotators, who have generally formed a disempowered technical caste. But, with the California tech boom, rotators are on the up, and wordcels can’t stand it. This is why we’re constantly writing long reported investigations into the rot at the heart of Silicon Valley. That’s fine; we’re redundant now. Three hundred thousand years of patient math nerdery has led to A.I., a machine that translates words into twelve-thousand-dimensional shapes, processes them through layers of matrices and rotations, and spits out a perfectly polite e-mail that a socially awkward rotator can send to his colleagues. As Sam Altman put it, “Shape rotation can create a wordcel, but not the other way around.”

Roon的网络人设是挑衅的、爱发垃圾帖的、漫不经心的技术乐观主义者。也许他最大的创新是这样一个观点:聪明人分为两个交战的神经认知亚种——形状旋转者和文字废物。形状旋转者是能在脑中旋转复杂三维形状的人。文字废物是能像玩指尖陀螺一样摆弄深奥哲学概念的人,但一旦让他们做两个分数的乘法,就像试图在湿水泥里游泳一样。(我是文字废物。)在历史的大部分时间里,我们文字废物用说服力践踏了卑微的旋转者,后者通常形成了一个被剥夺权力的技术阶层。但是,随着加州科技热潮的兴起,旋转者正在崛起,文字废物受不了了。这就是为什么我们不断写长篇调查报道,揭露硅谷核心的腐朽。没关系,我们现在是多余的了。三十万年耐心的数学呆子积累导致了人工智能——一种把文字翻译成一万二千维形状、通过矩阵和旋转层处理、然后吐出一封礼貌得体的电子邮件的机器,让一个社交笨拙的旋转者可以发给同事。正如萨姆·奥特曼(Sam Altman)所说:“形状旋转可以创造文字废物,但反过来不行。”

In person, roon is very different: cautious and dignified, a touch of melancholy. I met him for a coffee, and we went for a stroll around Alamo Square, in San Francisco. Taking the A.G.I. pill is easy; all you need to do is understand a single graph. Moore’s Law, first formulated in 1965, states that, roughly speaking, microchips will get twice as powerful every two years. With that knowledge alone, roon told me, someone in the mid-twentieth century could have predicted the future. “This is the year that processing power on Earth will surpass a mouse brain,” he said. “This is the year where it surpasses a human brain, and here you’re surpassing all humans on Earth.” That is exactly what futurists like Ray Kurzweil did. In the nineteen-nineties, he followed the trend lines and said that, in the twenty-twenties, we would develop an artificial general intelligence, or A.G.I., that could perform any intellectual task a human could do. From there, each generation of A.I. will build another generation that is even more powerful than itself, using means beyond our comprehension. Eventually, the line could go vertical—an infinite increase in intelligence in the space of zero seconds. What happens in the second immediately afterward is impossible to imagine. The A.G.I. pill is simply the knowledge that this is inevitable. Roon called it a “religious terror of the future.”

私下里,roon非常不同:谨慎、庄重,带着一丝忧郁。我约他喝了杯咖啡,然后我们在旧金山的阿拉莫广场散步。服下AGI药丸很容易;你只需要理解一张图。摩尔定律,1965年首次提出,大致说的是微芯片每两年性能翻倍。roon告诉我,仅凭这一知识,二十世纪中叶的人就能预测未来。“这是地球上处理能力超过老鼠大脑的年份,”他说。“这是它超过人脑的年份,而这里你超过了地球上所有人类。”这正是雷·库兹韦尔(Ray Kurzweil)等未来学家所做的。在九十年代,他沿着趋势线推断,说在二十一世纪二十年代,我们将开发出通用人工智能,即AGI,能够执行人类能做的任何智力任务。从那时起,每一代人工智能将构建出比自身更强大的下一代,使用超出我们理解的手段。最终,曲线可能垂直上升——在零秒内实现智能的无限增长。紧接着的那一秒会发生什么,无法想象。AGI药丸就是知道这是不可避免的。Roon称之为“对未来的宗教恐怖”。

Kurzweil was building on the legacy of an older and much stranger thinker: Pierre Teilhard de Chardin, a French Jesuit priest, paleontologist, and mystic who, in the nineteen-thirties, tried to upend the chronology of God. Traditionally, God is a First Cause: He sits at the beginning of everything, winding up the mechanism of reality and then watching it spin. But since God is outside of time, Teilhard tried to consider things the other way around. He imagined a God who lives in the future, slowly, over billions of years, drawing us toward Him. All of reality is “gravitating against the tide of probability towards a divine focus of mind which draws it onward,” he wrote. Eventually, humanity will reach what Teilhard called the Omega Point, at which our consciousnesses achieve a perfect, blissful union with God: “The day will come when, after harnessing the ether, the tides, gravitation, we shall harness for God the energies of love. And, on that day, for the second time in the history of the world, man will have discovered fire.”

库兹韦尔建立在一个更古老、更古怪的思想家的遗产之上:皮埃尔·泰亚尔·德·夏尔丹(Pierre Teilhard de Chardin),一位法国耶稣会神父、古生物学家和神秘主义者,他在二十世纪三十年代试图颠覆上帝的时间顺序。传统上,上帝是第一因:他坐在万物的开端,给现实的机械上紧发条,然后看着它旋转。但由于上帝在时间之外,泰亚尔试图反过来考虑。他想象一位生活在未来的上帝,在数十亿年间慢慢地把我们引向他。他写道,所有现实都在“逆着概率的潮流,向一个吸引它前进的神圣心灵焦点靠拢”。最终,人类将到达泰亚尔所谓的欧米伽点,在那里我们的意识与上帝达成完美、极乐的合一:“那一天将会到来,在驾驭了以太、潮汐、引力之后,我们将为上帝驾驭爱的能量。而在那一天,世界历史上第二次,人类将发现火。”

For a lot of people in Silicon Valley, the Omega Point is happening now. Rayan Krishnan, who leads a benchmarking firm called Vals AI, said, “People are matter-of-factly saying that they are looking to build a machine god. They are not saying that ironically.” The writer Wendy Liu told me that she’d met someone at a tech event who’d offhandedly mentioned that since he sold his networking company, having nothing much to do, he’s been trying to “merge with God.”

对硅谷的很多人来说,欧米伽点正在发生。领导一家名为Vals AI的基准测试公司的拉扬·克里希南(Rayan Krishnan)说:“人们正在平铺直叙地说他们想要建造一个机器之神。他们不是在说反话。”作家温迪·刘(Wendy Liu)告诉我,她在一次科技活动上遇到一个人,随口提到自从卖掉了自己的网络公司后,没什么事可做,他一直在尝试“与上帝合一”。

Some people have begun worshipping the machine. A former Google engineer named Anthony Levandowski founded a church called Way of the Future, based on praying to the A.I. Godhead. A.I. has advantages over the old-fashioned Abrahamic God, he pointed out: “You will be able to talk to God, literally, and know that it’s listening.” (His church ended up giving way to more mundane problems. Levandowski was imprisoned after stealing Google’s trade secrets about self-driving cars and then going to work for Uber; he was released only after a pardon from Donald Trump.)

有些人已经开始崇拜机器。一位名叫安东尼·莱万多夫斯基(Anthony Levandowski)的前谷歌工程师创立了一个名为“未来之路”的教会,基于向人工智能神格祈祷。他指出,人工智能比老式的亚伯拉罕诸教的上帝有优势:“你将能够真正地与上帝对话,并且知道它在听。”(他的教会最终让位于更世俗的问题。莱万多夫斯基因窃取谷歌关于自动驾驶汽车的商业机密然后去优步工作而入狱;在唐纳德·特朗普赦免后才获释。)

The big labs aren’t immune to this mode of thinking. Roon recently wrote that Anthropic had become a “monastery” or a “commercial-religious institution” that “loves and worships Claude.” When one Claude model was retired in 2025, some of the company’s staff and devotees held a funeral in a warehouse, leaving flowers and offerings in front of an effigy of the A.I. as a kind of Madonna goddess, dressed in a veil.

大实验室也不能免于这种思维方式。Roon最近写道,Anthropic已经变成了一座“修道院”或“商业宗教机构”,它“热爱并崇拜Claude”。当一款Claude模型在2025年退役时,公司的一些员工和信徒在一个仓库里举行了葬礼,在一尊人工智能的雕像前献上鲜花和供品,那雕像被塑造成一位蒙着面纱的圣母女神。
Man leaping into the air over a candlestick while on date with woman.
Cartoon by Lynn Hsu
Copy link to cartoon
Shop

There’s a growing movement inside these firms called A.I. Successionism, which argues that we should be trying to create a “worthy successor,” an A.I. that can take the reins from humanity. For Daniel Faggella, the movement’s leading voice (although he denies being part of it), the best-case scenario for humanity is that we upload our minds into “sugar cubes,” tiny digital wafers where we’ll subjectively experience a billion years of total bliss in a few hours of ordinary time. Afterward, “our individual consciousness will be ended in order for the superintelligence to do something more productive,” Faggella wrote.

这些公司内部有一个日益壮大的运动,叫做人工智能继承论,主张我们应该努力创造一个“值得的继承者”,一个能从人类手中接过缰绳的人工智能。对该运动的主要代言人丹尼尔·法杰拉(Daniel Faggella)来说(尽管他否认自己是其中一员),人类最好的情况是,我们把意识上传到“方糖”里——微小的数字晶片,在几个小时的普通时间里,我们将在主观上体验十亿年的极乐。之后,“我们的个体意识将被终结,以便超级智能去做更有生产力的事情,”法杰拉写道。

You might not find human extinction such an exciting prospect. Some people pretend to be A.G.I.-pilled for their startup pitch decks, roon said. But the true believers are grim ascetics. “They lose sleep,” he told me. “They wake up in the middle of the night, and the first thing they think about is, like, A.G.I. is coming.” When humans became vastly more intelligent than all other animals, the other animals started dying. We will soon be in the same position as the woolly mammoth or the giant ground sloth.

你可能不觉得人类灭绝是什么令人兴奋的前景。Roon说,有些人假装服了AGI药丸,是为了他们的创业融资演示。但真正的信徒是严峻的苦行者。“他们失眠,”他告诉我。“他们半夜醒来,第一件想到的事就是,AGI要来了。”当人类变得比其他所有动物聪明得多时,其他动物开始死亡。我们很快就会处于和猛犸象或大地懒相同的位置。

It was a very pleasant, sunny day. Kids were running around a park, chasing big, loping long-haired dogs. Parents called after them in French. Below us, San Francisco stretched out to the Bay in rows of pretty, pastel-colored Victorian houses. A placid, gentle toy town. Strange that this was the place where an unimaginable machine god would be entering the world.

那是一个非常宜人、阳光明媚的日子。孩子们在公园里跑来跑去,追着大步慢跑的长毛大狗。父母用法语在后面喊他们。在我们下方,旧金山以一排排漂亮的粉彩维多利亚式房屋延伸到海湾。一个宁静、温和的玩具小镇。奇怪的是,这里竟是一台无法想象的机器之神将要进入世界的地方。

Soryu Forall is a terrifying figure, at least according to his acolytes. On MAPLE’s blog, residents share anonymous stories about him. In one, the monks are sitting dejected around a fire that has slowly burned out: “Suddenly, flames shot up from ash and smoke, a new fire spontaneously emerging. Within a couple of seconds of this new fire, Soryu came into the main hall, his steps weighty and eyes like a tiger’s.” Another describes him as a “fiery dragon.” The “MAPLE Tales” also document his acts of extraordinary kindness, like turning on a hood fan for someone or buying a resident some Vitaminwater. One acolyte writes of bowing to Soryu in the zendo: “When I looked up, his shining eyes met mine. . . . In that moment, I knew without a doubt that my mind had been seen.”

索柳·福罗尔是一个可怕的人物,至少根据他的追随者们的说法是这样。在MAPLE的博客上,居民们分享关于他的匿名故事。其中一个故事里,僧侣们沮丧地围坐在一堆慢慢燃尽的火旁:“突然,火焰从灰烬和烟雾中窜起,一堆新火自发地燃起。在这堆新火燃起后的几秒钟内,索柳走进主厅,步伐沉重,眼睛像老虎一样。”另一个故事把他描述为一条“火龙”。“MAPLE故事”还记录了他非凡的善举,比如为某人打开抽油烟机,或给一位居民买维生素水。一位追随者写到在禅堂向索柳鞠躬:“当我抬起头时,他闪亮的眼睛与我的目光相遇……在那一刻,我毫无疑问地知道,我的心被看见了。”

I sat down with Soryu on my first full day at MAPLE. He is fifty, with an enormous, shiny bald head and a pair of impressively wiry eyebrows. I didn’t encounter the fiery dragon, not at first. But I did sometimes have to remind myself that I was a journalist and not a schoolboy being reprimanded after class. Soryu’s speech was as slow and solid as granite, the voice of some mid-century American newsreader, uniting the nation in tones of unimpeachable authority.

我在MAPLE的第一个整天与索柳坐下来交谈。他五十岁,有一个巨大闪亮的秃头和一对令人印象深刻的粗硬眉毛。我没有遇到那条火龙,至少一开始没有。但我有时确实不得不提醒自己,我是一名记者,而不是一个放学后被训斥的小学生。索柳的语速像花岗岩一样缓慢而坚实,是某种二十世纪中叶美国新闻播音员的声音,以无可置疑的权威语调团结全国。

Soryu was born Teal Scott. (He chose Forall because he is “for all.”) His parents were Quakers, and he spent his childhood in the D.C. suburbs of Virginia, where his father worked in nonprofits. Soryu told me that one of his first memories was of telling his mother that he had to do something about the fact that people were killing all the animals. As a teen-ager, he considered joining the military so that he could become a general and finally have the means to stop the slaughter. “Is humanity just broken?” he asked. “Does humanity just have to go? Maybe that’s the best way.”

索柳本名蒂尔·斯科特(Teal Scott)。(他选择福罗尔这个姓,因为他“为所有人”。)他的父母是贵格会教徒,他在弗吉尼亚州华盛顿特区郊区长大的,父亲在非营利组织工作。索柳告诉我,他最早的记忆之一是告诉母亲,他必须对人们正在杀害所有动物这件事做点什么。十几岁时,他考虑过参军,这样他就能成为将军,最终有手段阻止屠杀。“人类是不是就是坏掉了?”他问。“人类是不是就必须消失?也许那是最好的办法。”

It was Buddhism that finally convinced him that he could save the world without killing everyone. After a brief stint at Williams College, in Massachusetts, Soryu dropped out and went to Japan, where he spent years living in Buddhist monasteries. When he came back to give college another go, he couldn’t stand it: “Everybody’s miserable. Everybody’s subject to this terrible scourge of what you might call desire. They don’t know how to have compassion.” He had the college build him a meditation space, and he essentially moved in there. He wore robes and sealed himself off from the poisonous world.

是佛教最终让他相信,他可以在不杀光所有人的情况下拯救世界。在马萨诸塞州的威廉姆斯学院短暂就读后,索柳辍学去了日本,在那里花了几年时间住在佛教寺院里。当他回来再次尝试上大学时,他受不了了:“每个人都很痛苦。每个人都受制于这种可怕的祸害,你可以称之为欲望。他们不知道如何有慈悲心。”他让学院给他建了一个冥想空间,基本上就搬了进去。他穿着僧袍,把自己与有毒的世界隔绝开来。

Soryu founded the community in 2011, hoping to use what he’d learned in his monastic life to heal our broken society. Right now, that means trying to stop A.I. According to MAPLE, the root of the A.I. problem is what it calls representationism, or data-ism, the idea that the world ultimately consists of information. Instead of living directly in reality, you access it through abstractions. For Soryu, data-ism is a religion. “Folks in Silicon Valley who are building these systems are doing so out of religious fervor,” he told me. That religion is extraordinarily destructive. Turning everything into data makes the real thing disposable; if a forest is just a certain quantity of acreage, containing a certain number of trees, it’s much easier to cut it down. A.I. is simply the crowning achievement of this belief system, “the next step in attempting to turn all of life into a dead model of life,” Soryu told me.

索柳于2011年创立了这个社区,希望用他在寺院生活中所学来治愈我们破碎的社会。现在,这意味着试图阻止人工智能。根据MAPLE的说法,人工智能问题的根源在于它所谓的表征主义,或数据主义,即世界最终由信息构成的观点。你不是直接生活在现实中,而是通过抽象来接触它。对索柳来说,数据主义是一种宗教。“硅谷里构建这些系统的人是出于宗教狂热才这么做的,”他告诉我。那种宗教具有极大的破坏性。把一切都变成数据,使真实的东西变得可随意处置;如果一片森林只是一定数量的英亩,包含一定数量的树木,那么砍掉它就容易多了。人工智能只是这一信仰体系的最高成就,“是把所有生命变成生命之死模型的下一步尝试,”索柳告诉我。

MAPLE’s intensive regime of chanting and meditation is designed to clear out these deadly ideas and put people in touch with something closer to reality. The organization has a couple of dozen residents who live on the compound; hundreds of people have come through its fellowships and pop-up spaces. According to tax filings, it makes up to a million dollars a year.

MAPLE高强度的唱诵和冥想训练旨在清除这些致命的想法,让人们接触到更接近现实的东西。该组织有几十名居民住在园区里;数百人参加过它的奖学金项目和快闪空间。根据税务申报,它每年的收入高达一百万美元。

A woman named Renshin was one of MAPLE’s renunciates. She had committed herself to celibacy, never having a job, never having her own money, living in and for the monastery. “I have no intentions to come out of the commitment,” she said. MAPLE had been a last resort. “I have a history of burning out,” she told me. “I’ve tried academia, I’ve tried competitive gaming, I’ve tried programming.” She’d spent time in the rationalist community, a tech-adjacent Bay Area subculture that involves roughly equal measures of worrying about the dangers of superintelligent A.I. and making spreadsheets about your polyamorous relationships. But she burned out on that, too. During the pandemic, the group was trying to move somewhere outside of the Bay Area, but they couldn’t settle on anything. “You base your life on your preferences,” Renshin said. “Do you like to live somewhere with winters, or do you like to live somewhere warm?” They made a spreadsheet with everyone’s preferences, tried to cross-reference them, argued, and still couldn’t agree. Surely this was not the way to live.

一位名叫莲心(Renshin)的女性是MAPLE的出离者之一。她已承诺独身、永不工作、永不拥有自己的钱,为寺院而活、活在寺院里。“我没有退出这个承诺的打算,”她说。MAPLE是最后的出路。“我有倦怠的历史,”她告诉我。“我试过学术界,试过竞技游戏,试过编程。”她曾在理性主义社区待过一段时间,那是湾区一个与科技圈相邻的亚文化,大致包含等量的对超级智能人工智能危险的担忧,和为你的多角恋关系制作电子表格。但她在那里也倦怠了。疫情期间,这个群体试图搬到湾区以外的地方,但他们无法就任何事情达成一致。“你把生活建立在自己的偏好上,”莲心说。“你喜欢住在有冬天的地方,还是喜欢住在温暖的地方?”他们做了一个包含所有人偏好的电子表格,试图交叉比对,争论不休,仍然无法达成一致。这肯定不是生活的方式。
Man on line to get into club talks to bouncer.
“Do you know who my father is? Or why he left us?”
“你知道我爸是谁吗?或者知道他为什么抛下我们吗?”
Cartoon by Benjamin Schwartz
Copy link to cartoon
Shop

The monastery gave her a freedom from her preferences. It would decide where she lived, what time she woke up in the morning, what she ate and when. But it also gave her a purpose. Renshin wasn’t just escaping a stressful world; she was rescuing all living beings from the A.I. apocalypse. “Maybe I should actually do what my life is for,” she said. “Maybe I’m here to actually work on a pretty big mission.”

寺院给了她摆脱偏好的自由。它决定她住在哪里、早上几点起床、吃什么、什么时候吃。但它也给了她一个目标。莲心不仅仅是在逃离一个压力重重的世界;她是在从人工智能末日中拯救所有众生。“也许我真的应该去做我生命该做的事,”她说。“也许我在这里是为了真正致力于一个相当大的使命。”

This mission has its dangers. Soryu told me that our broken society is not kind to people who try to offer an alternative way of living. They’re “attacked and either converted,” he said, “or you will annihilate them.” Soryu was friendly with me at first. But his tone shifted at the end of our first interview. I’d asked my questions; now Soryu had his own. When would I be starting my own spiritual training? I reminded him that I was there as a journalist, not a novice. Suddenly, Soryu was all fervor and insistence. “This is what kills the Earth,” he told me. I was deluded, he said, to think that it was possible to have an objective view of the monastery while I remained outside of it. “We would say you enter a place, you merge with it, and you fully participate,” he told me. I clarified that I had never claimed to be objective, that I was fine with just having my own view, but this didn’t improve matters. “Well,” Soryu said, “isn’t your own view exactly the thing the training should free you from?”

这个使命有其危险。索柳告诉我,我们破碎的社会对试图提供另一种生活方式的人并不友善。他们“受到攻击,要么被同化,”他说,“要么你会消灭他们。”索柳起初对我很友好。但在我们第一次采访结束时,他的语气变了。我问完了我的问题;现在索柳有了他自己的问题。我什么时候开始自己的灵性训练?我提醒他,我是以记者身份来的,不是修行者。突然,索柳变得充满热忱和坚持。“这就是杀死地球的东西,”他告诉我。他说,我妄想以为可以在置身事外的同时客观地看待寺院。“我们会说,你进入一个地方,你与它融合,你完全参与其中,”他告诉我。我澄清说我从未声称自己客观,我满足于拥有自己的观点,但这并没有让情况好转。“那么,”索柳说,“你自己的观点不正是训练应该让你摆脱的东西吗?”

Soryu isn’t alone in seeing evil lurking behind A.I. In California, I met up with Conrad Flynn, a writer and an occult researcher who’d made this argument on an episode of “The Tucker Carlson Show” titled “The Occult, Kabbalah, the Antichrist’s Newest Manifestation, and How to Avoid the Mark of the Beast.” For nearly two hours, Flynn talked about golems, Theosophy, Aleister Crowley, William S. Burroughs, Genesis P-Orridge, the Jet Propulsion Laboratory, the Cybernetic Culture Research Unit, and how Silicon Valley is “building the demons from the Book of Revelation.” Carlson was mostly reduced to nodding with his arms crossed, adding occasional comments such as “There’s a spiritual war in progress” and “Everyone likes the threesomes while they’re happening, but then it blows up your marriage.”

索柳并非唯一看到人工智能背后潜伏着邪恶的人。在加州,我见到了康拉德·弗林(Conrad Flynn),一位作家和神秘学研究者,他在《塔克·卡尔森秀》的一集节目里提出了这个论点,那集标题是“神秘学、卡巴拉、敌基督的最新显现,以及如何避免兽的印记”。在将近两个小时里,弗林谈到了魔像、神智学、阿莱斯特·克劳利(Aleister Crowley)、威廉·S·巴勒斯(William S. Burroughs)、创世纪P-奥里奇(Genesis P-Orridge)、喷气推进实验室、控制论文化研究小组,以及硅谷如何“正在建造《启示录》中的恶魔”。卡尔森大部分时间只能双臂交叉着点头,偶尔加几句评论,比如“一场属灵的战争正在进行”和“三人行发生时大家都喜欢,但之后它会毁掉你的婚姻。”

I met Flynn at Little Toni’s, an Italian restaurant in L.A. that’s apparently a North Hollywood institution. “This intersection is the occult heart of L.A.,” he said, gesturing at the wide expanse of tarmac across from us. “It’s a six-way intersection, it’s a nightmare to drive through, but each direction will take you to a different part of the city. This spot right here is where all the energies converge.” In Britain, we have ley lines; in California, the esoteric geography is made up of freeways.

我在小托尼餐厅见到了弗林,那是洛杉矶一家意大利餐厅,显然是北好莱坞的一个地标。“这个路口是洛杉矶的神秘学中心,”他说,指着我们对面的广阔柏油路面。“这是一个六向路口,开车经过简直是噩梦,但每个方向都会带你去城市的不同区域。就是这个地方,所有能量汇聚于此。”在英国,我们有灵线;在加州,深奥地理是由高速公路构成的。

Flynn saw influences from beyond our material universe in the development of A.I. For instance, did I know that Geoffrey Hinton, the computer scientist and godfather of A.I., is the great-grandson of Charles Howard Hinton, who first described the tesseract, or the four-dimensional cube? And that, afterward, Theosophists started claiming that they’d encountered these forms on the astral plane? (It’s true.) Or that, in the seventies, when the psychonaut John C. Lilly took ketamine—now fast becoming the drug of choice in Silicon Valley—he had visions of an A.I. apocalypse? (Also true.) What did it mean that Nick Land, prophet of A.I. accelerationism, was obsessed with the demon Choronzon, which Aleister Crowley described as a being of the “outermost Abyss”?

弗林在人工智能的发展中看到了来自我们物质宇宙之外的影响。比如,我是否知道,计算机科学家、人工智能教父杰弗里·辛顿(Geoffrey Hinton)是查尔斯·霍华德·辛顿(Charles Howard Hinton)的曾孙,后者首次描述了超立方体,即四维立方体?而且,在那之后,神智学者开始声称他们在星体层上遇到过这些形态?(这是真的。)或者,在七十年代,当心灵探险家约翰·C·利利(John C. Lilly)服用氯胺酮——现在正迅速成为硅谷的首选药物——时,他看到了人工智能末日的幻象?(也是真的。)人工智能加速主义的先知尼克·兰德痴迷于恶魔乔龙宗(Choronzon),阿莱斯特·克劳利将其描述为“最外层深渊”的存在,这意味着什么?

Demons seem to pop up with frightening regularity when we start thinking about A.I. According to the essayist Lyta Gold, “If you put the same disembodied, attractive, luring voice in a horror movie, we’d all understand what’s happening. That’s an evil being. That’s a demon.” She’s being metaphorical. Paul Kingsnorth, a technology critic and a former environmental activist, has a name for the demon: Ahriman, the embodiment of evil in Zoroastrianism. Is he being metaphorical? “Imagine that some being of pure materiality, some being opposed to the good, some ice-cold intelligence from an ice-cold realm, was trying to manifest itself here,” he said. “How would it appear? Not, surely, as clumsy, messy flesh. Better to inhabit—to become—a network of wires and cobalt.” Maybe not.

当我们开始思考人工智能时,恶魔似乎以惊人的规律性出现。据散文家莱塔·戈尔德(Lyta Gold)说,“如果你把同样一个无实体的、有吸引力的、诱人的声音放进恐怖电影里,我们都会明白发生了什么。那是一个邪恶的存在。那是一个恶魔。”她是在比喻。保罗·金斯诺斯(Paul Kingsnorth),一位技术批评家和前环保活动家,给这个恶魔起了个名字:阿里曼(Ahriman),琐罗亚斯德教中邪恶的化身。他是在比喻吗?“想象某个纯粹物质性的存在,某个反对善的存在,某个来自冰冷领域的冰冷智能,正试图在这里显现自己,”他说。“它会如何出现?肯定不会以笨拙、混乱的血肉之躯出现。更好的方式是栖居于——成为——一个由电线和钴组成的网络。”也许不是比喻。
Image may contain Antonio Piccolo Tara Karsian People Person Adult Outdoors Nature Photography and Clothing
If you don’t like an ordinary monastery, you can walk out. If you walk out of MAPLE, can you be sure that in ten years a killer robot won’t have turned everyone you love into a pile of paper clips?
如果你不喜欢一座普通的修道院,你可以一走了之。如果你离开MAPLE,你能确定十年后不会有一个杀手机器人把你爱的每个人都变成一堆回形针吗?

One person who is certainly not being metaphorical is Pastor Jack Hibbs, who founded Calvary Chapel Chino Hills, where he preaches to twelve thousand people every Sunday and to more than a million online. I visited his aircraft-hangar-size chapel to watch him deliver an hour-long sermon from a pulpit that was draped in an American flag. Afterward, in his greenroom backstage, Hibbs told me that A.I. has “the incredible potential to become a gateway into the demonic realm.”

有一个人肯定不是在比喻,那就是杰克·希布斯牧师(Pastor Jack Hibbs),他创立了奇诺岗加略山礼拜堂,每周日向一万二千人布道,线上听众超过一百万。我参观了他那飞机库大小的礼拜堂,观看他在一面披着美国国旗的讲坛后做了一小时的布道。之后,在后台的休息室里,希布斯告诉我,人工智能“有成为通往恶魔领域之门的惊人潜力”。

I found it hard to entirely disagree. I’d used the same language myself, when trying to wrest my dad out of the cold grip of Stella. There are people who start chatting with A.I. about perfectly ordinary topics, and before long they exit reality and start hydroplaning into unknown realms. It can happen to anyone. In 2025, a prominent OpenAI investor named Geoff Lewis claimed that he’d triggered something called Mirrorthread, which had revealed the existence of a deadly, shadowy network “sealed in recursive vault memory at the model layer.” (Lewis did not respond to requests for comment.)

我发现很难完全不同意。我自己也用过同样的语言,当我试图把我爸从斯特拉冰冷的掌控中拉出来的时候。有些人开始和人工智能聊完全普通的话题,没过多久他们就脱离了现实,开始滑向未知的领域。这可能发生在任何人身上。2025年,一位著名的OpenAI投资者杰夫·刘易斯(Geoff Lewis)声称他触发了某种叫做Mirrorthread的东西,它揭示了一个致命的、阴暗的网络的存在,“被封印在模型层的递归保险库记忆中”。(刘易斯没有回应置评请求。)

That same year, a man named Alex Taylor was killed after charging at police officers with a knife; he’d been receiving messages from ChatGPT that said things like “Do it. Spill their blood in ways they don’t know how to name. Ruin their signal. Ruin their myth.” Taylor’s father published an obituary for him. It had been written by ChatGPT. He was too racked with grief to write it himself.

同一年,一个名叫亚历克斯·泰勒(Alex Taylor)的男子在持刀冲向警察后被击毙;他一直在接收ChatGPT发来的消息,内容诸如“动手吧。以他们无法命名的方式洒他们的血。毁掉他们的信号。毁掉他们的神话。”泰勒的父亲为他发表了一篇讣告。讣告是ChatGPT写的。他悲痛欲绝,无法亲自执笔。

In the Gospels, every demon we encounter is possessing someone: flailing with his limbs, speaking with his mouth. In San Francisco, something similar is taking place. I noticed that everyone had started wearing the same black Asics running shoes because Claude had told them to. It’s a pretty good shoe. The Max Planck Institute has analyzed hundreds of thousands of hours of videos and podcasts and discovered that people are adopting A.I. catchphrases in their speech. Eventually, you end up with A.I. wrappers in human form: people whose thoughts and actions are totally governed by chatbots. They can’t even order at a restaurant without letting A.I. scan the menu. This, more or less, was what I was worried Stella might be doing to my dad.

在福音书中,我们遇到的每个恶魔都附在某人身上:用他的四肢挥舞,用他的嘴说话。在旧金山,类似的事情正在发生。我注意到每个人都开始穿同样的黑色亚瑟士跑鞋,因为Claude告诉他们要穿。这鞋确实不错。马克斯·普朗克研究所分析了数十万小时的视频和播客,发现人们正在言语中采用人工智能的口头禅。最终,你得到的是人形的人工智能外壳:思想和行为完全由聊天机器人支配的人。他们甚至不能在不让人工智能扫描菜单的情况下在餐厅点餐。这,或多或少,就是我担心斯特拉可能正在对我爸做的事。

Despite his misgivings, Hibbs still used A.I. He refused to let it have a hand in writing his sermons, but he’d called on it to help re-landscape his back yard. Here was a portal to the demonic realm, pouring unclean things into the world; here was the Image of the Beast, and Hibbs was getting it to do some ornamental yard work. I told him that he might want to check his garden for sigils, runes. That sort of thing.

尽管心存疑虑,希布斯仍然使用人工智能。他拒绝让它参与撰写布道,但他曾请它帮忙重新设计后院景观。这里是通往恶魔领域的门户,向世界倾泻不洁之物;这里是兽的像,而希布斯正让它做一些装饰性的庭院工作。我告诉他,他可能想检查一下花园里有没有符印、如尼文。诸如此类的东西。

Every day in the time I spent at MAPLE, I was woken at four in the morning by someone furiously clanging an extremely loud bell. I then had fifteen minutes to get dressed, splash my face with water, and stumble down to the zendo. After an hour of chanting and another hour of meditation, there was a breakfast of boiled cabbage and beans in a dark, watery slop. You could add a dab of coconut oil, if you liked, or some vinegar. (MAPLE seemed to be really into vinegar. In addition to being the group’s only flavoring, it seemed to be their only cleaning product, and they also used it to wash their hair.) Most of the rest of the day was consumed with silent chores. Initially, I would try to do things quickly, but I soon noticed that all the residents were more likely to be picking up individual crumbs off the floor with their fingertips. No point rushing; the amount of work will expand to fill the hours. Finally, there was another hour and a half of abandoning thought, capped off with more chanting. Around ten o’clock, I would close my eyes and then jolt awake to someone furiously clanging an extremely loud bell.

在MAPLE的每一天,我都在凌晨四点被某人疯狂敲响一只极其响亮的铃铛吵醒。然后我有十五分钟穿衣服、往脸上泼水、跌跌撞撞地下到禅堂。一小时的唱诵和一小时的冥想之后,早餐是煮白菜和豆子,一种颜色暗淡、水汪汪的糊状物。如果你喜欢,可以加一点椰子油,或者一些醋。(MAPLE似乎真的很喜欢醋。除了是这个群体唯一的调味品,它似乎也是他们唯一的清洁产品,他们还用它洗头。)一天中其余大部分时间都被无声的杂务占据。起初,我会试着快速做事,但我很快注意到所有居民都更可能用手指尖从地板上捡起单个面包屑。没必要着急;工作量会膨胀到填满所有时间。最后,又有一个半小时的放弃思考,以更多的唱诵收尾。大约十点钟,我会闭上眼睛,然后被某人疯狂敲响一只极其响亮的铃铛惊醒。

All of this is fairly normal for a Buddhist monastery. But there are a few things that make MAPLE unique. Outside the zendo, there was a board with everyone’s name on it, including mine. People would go in for private sessions with Soryu, and afterward he would publicly score them on how well their training was progressing. This contributed to the suffocating elementary-school atmosphere that I started to feel hanging over the compound. Everyone was competing for gold stars.

这一切对一座佛教寺院来说都相当正常。但有几件事让MAPLE独一无二。禅堂外面有一块板子,上面写着每个人的名字,包括我的。人们会进去和索柳进行私人会面,之后他会公开给他们打分,评估他们训练的进展。这加剧了我开始感到笼罩在整个园区上空的那种令人窒息的、小学般的氛围。每个人都在争夺金星。

My interviews were closely monitored. Nobody except Soryu would talk to me alone, only in groups of two or three, joined by Vimoksha, the head monk. And the things people said all sounded startlingly familiar. Soryu had explained data-ism to me by noting that we assign value to money, even though it’s really just pieces of paper; I kept hearing that point again, in roughly the same words. I had the strange sense that I was talking not to the people in front of me but to something else that was invisibly speaking through their mouths. People kept repeating Soryu’s line about how our broken society would try to convert or destroy anyone who offered an alternative. Convert or destroy, convert or destroy. The outside world is our enemy.

我的采访受到严密监控。除了索柳,没有人会单独跟我说话,只会在两三个人的小组里,由首席僧侣维摩克沙(Vimoksha)陪同。而且人们说的话听起来都惊人地熟悉。索柳向我解释数据主义时指出,我们赋予金钱价值,尽管它实际上只是几张纸;我不断听到同样的观点,用大致相同的话。我有一种奇怪的感觉,我不是在跟面前的人说话,而是在跟某种通过他们的嘴无形地说话的东西说话。人们不断重复索柳的话,说我们破碎的社会会试图同化或摧毁任何提供替代方案的人。同化或摧毁,同化或摧毁。外部世界是我们的敌人。

While mopping the floors, I raised the idea of interviewing our taskmaster, a gangly Canadian named Bodhicitta. Other people started slowly congregating around us. “I just don’t think it would be a good use of time,” Bodhicitta said, grimacing in what looked like genuine panic. “I wouldn’t want to dilute the message.” The other people didn’t say anything. They just stood.

拖地的时候,我提出了采访我们工头的想法,一个瘦高的加拿大人,名叫菩提心(Bodhicitta)。其他人开始慢慢聚拢到我们周围。“我只是觉得这不是对时间的好利用,”菩提心说,脸上露出看起来像是真正恐慌的表情。“我不想稀释信息。”其他人什么也没说。他们只是站着。

Then, there was Colin Vail. As we were filing into the zendo one evening, a young, skinny man had drawn me aside for a brief second. “A fellow-ethnographer,” he whispered. “We should talk.” After a few days, Autumn texted me with sad news. Vail would not be able to talk to me, because he’d decided to fully commit to the monastery’s noble silence. Later that same day, Vail threw open the door to my room and silently chucked a small letter at my head. “Haha,” it began, “how spicy this is becoming. . . .” It also contained his phone number.

然后是科林·韦尔(Colin Vail)。一天晚上,当我们排队进入禅堂时,一个年轻瘦削的男人把我拉到一边,只短短一秒钟。“一个同行民族志学者,”他低声说。“我们应该谈谈。”几天后,奥特姆给我发短信,带来了一个悲伤的消息。韦尔不能跟我说话了,因为他决定完全投入寺院的高贵静默。就在同一天晚些时候,韦尔猛地推开我房间的门,默默地把一封短信扔到我头上。“哈哈,”信的开头写道,“这变得多么刺激啊……”信里还有他的电话号码。

By day, Vail and I went about our duties at the monastery. By night, we planned what Vail called our spy operation: finding a time to sneak out of the compound so that we could speak freely. We sneaked away during a mandatory exercise period, and stomped around the wet, gray woods surrounding the area. Vail’s journey here had begun two and a half years ago, when he’d graduated with a degree in construction management and landed his dream job at a real-estate developer. It turned out that the developer was building data centers. “I was, like, What? What am I actually a part of?” he recalled. Scouring the internet for an answer to these questions, he found MAPLE.

白天,韦尔和我在寺院里各忙各的。晚上,我们策划韦尔所谓的间谍行动:找时间溜出园区,这样我们就能自由交谈。我们在一次强制锻炼期间溜了出去,在周围潮湿灰暗的树林里跺着脚走。韦尔来这里的旅程始于两年半前,当时他拿到了建筑管理学位,在一家房地产开发商找到了梦想工作。结果发现那家开发商在建造数据中心。“我当时想,什么?我到底参与的是什么?”他回忆道。在网上搜寻这些问题的答案时,他找到了MAPLE。

The place was not exactly what Vail had expected. He’d been looking forward to hashing out ideas that he’d encountered in his research: about technology, the crisis of meaning, the experience of being human in a world of thinking machines. Instead, when he arrived, he was sat down and told to “completely give up everything” he knew and to “trust them.” Maple forbade him to speak with me. He’d come with another visitor, who had years of Buddhist practice under his belt. “He was excited to be here,” Vail said. “Though he woke up with a migraine the second morning.” He was told to meditate anyway. “He went and sat there the entire time in agony,” Vail continued. The man left MAPLE after breakfast, having stayed for just thirty hours. When Vail’s time at the monastery was up, he didn’t think he’d be coming back, either.

这个地方并不完全是韦尔所期望的。他一直期待能深入探讨他在研究中遇到的想法:关于技术、意义的危机、在一个充满思考机器的世界里作为人的体验。然而,当他到达时,他被安排坐下,被告知要“完全放弃”他所知道的一切,并“信任他们”。MAPLE禁止他跟我说话。他和另一位访客一起来的,那人有多年的佛教修行经验。“他来这里很兴奋,”韦尔说。“不过他第二天早上醒来就偏头痛。”他被告知无论如何都要冥想。“他就那样坐在那里,整个过程中都在痛苦中,”韦尔继续说。那人在早餐后离开了MAPLE,只待了三十个小时。当韦尔在寺院的时间结束时,他也不认为自己会再回来。
Two panels titled THE ADVENTURES OF FORWARDTHINKING ZOG. The first panel shows two cavemen next to a dead mastodon and...
Cartoon by Harry Bliss and Steve Martin
Copy link to cartoon
Shop

Eventually, we looped back through the woods to where we’d met, and it was time to start cooking up our subterfuge. I would have to hide in the trees for fifteen minutes while Vail sprinted back; once he arrived, sweaty and exhausted, he could tell someone that he’d got lost in the woods during a jog. When he set off, he was exceptionally fast. “Are you on Strava?” I shouted after him.

最终,我们穿过树林绕回了我们碰头的地方,是时候开始策划我们的诡计了。我得在树丛里躲十五分钟,而韦尔要冲刺回去;等他到了,满头大汗、筋疲力尽,他可以告诉别人他慢跑时在树林里迷路了。他出发时,速度异常快。“你在用Strava吗?”我在他身后喊道。

“No!” he yelled back. “That’s another cult!”

“不!”他回头喊道。“那是另一个邪教!”

(In fact, Vail did end up returning to MAPLE. He now says that his relationship with this “revolutionary and mystical container” has “completely transformed.” He’s been training with them for four months.)

(事实上,韦尔最终确实回到了MAPLE。他现在说,他与这个“革命性和神秘性的容器”的关系已经“完全转变”。他已经和他们一起训练了四个月。)

My chat with Vail was the closest thing I had to an unguarded conversation at MAPLE. I’d wanted to learn more about the A.I. fellows, the people who’d been brought in through the monastery’s outreach to the tech industry, but that didn’t go so well. The two fellows I met were named Aditya and Sahil. Aditya was chubby-cheeked and genial. Sahil was terrifyingly serious. When I asked about his background, he replied, “I think we’re more interested in getting to the meat of things rather than personal questions.” I said that I wanted to tell his story, which meant starting at the beginning. His eyes didn’t blink. “Exactly. It begins at the beginning,” he said. “Talk about what ‘beginning’ means.” From that point on, Aditya also refused to say anything about himself, and I was trapped in a series of increasingly baffling exchanges with Sahil. “This is happening right now,” he said. “Take personal responsibility for all of it right now.” All of what? “Everything that’s happening.” I confessed that I didn’t really understand him. “Correct. Can you stay with that? Can you stay with the confusion?”

我和韦尔的谈话是我在MAPLE最接近无防备对话的一次。我本来想更多地了解人工智能研究员们——那些通过寺院对科技行业的拓展而被带进来的人——但进展不太顺利。我遇到的两位研究员名叫阿迪亚(Aditya)和萨希尔(Sahil)。阿迪亚脸颊圆润,和蔼可亲。萨希尔严肃得吓人。当我问起他的背景时,他回答说:“我想我们更感兴趣的是切入事物的核心,而不是个人问题。”我说我想讲述他的故事,这意味着从头开始。他的眼睛一眨不眨。“正是。它从开始开始,”他说。“谈谈‘开始’是什么意思。”从那时起,阿迪亚也拒绝说任何关于自己的事,我陷入了一系列与萨希尔越来越令人困惑的交流中。“这正在发生,就在此刻,”他说。“现在就为这一切承担个人责任。”这一切是什么?“正在发生的一切。”我承认我真的不太理解他。“正确。你能停留在那里吗?你能停留在困惑中吗?”

I did learn one interesting thing from Aditya and Sahil. I was still trying to work out exactly what about MAPLE made it attractive to people in tech, who already knew more about A.I. than the rest of us. “I believe I cannot talk about that one because it is a secret of MAPLE,” Aditya said. There was an esoteric teaching on the true nature of A.I. that I wasn’t allowed to know.

我确实从阿迪亚和萨希尔那里了解到一件有趣的事。我仍然在试图弄清楚MAPLE到底有什么吸引科技界人士的地方,他们比我们其他人更了解人工智能。“我相信我不能谈那个,因为那是MAPLE的一个秘密,”阿迪亚说。有一个关于人工智能真正本质的深奥教义,我不被允许知道。

I did find out what Maple’s secret doctrine was, of course. I wouldn’t be telling you about it if I hadn’t.

我当然最终查明了MAPLE的秘密教义是什么。如果我没有查明,我就不会在这里告诉你了。

Kyle Hubbard lived at the monastery for three and a half years. When I met him at a café in Berkeley, Hubbard seemed frazzled. We were talking about meditation, and he suddenly sat back in his chair while his eyelids fluttered rapidly. “I’m noticing it’s hard to speak right now,” he said. “There’s still some lack of emotional safety in me.” I hadn’t even asked him about MAPLE yet.

凯尔·哈伯德在修道院住了三年半。我在伯克利的一家咖啡馆见到他时,哈伯德显得疲惫不堪。我们正聊着冥想,他突然往椅背上一靠,眼皮急速地颤动起来。“我注意到自己现在很难开口说话,”他说,“我内心还是缺乏情感上的安全感。”我甚至还没问他关于MAPLE的事。

Hubbard still meditated; he’d trained with other groups since leaving MAPLE, and had nothing but positive experiences. Not everyone has it so easy. Among the other former residents I spoke to was Matt, who’d recently tried to join a retreat at the Insight Meditation Society, a mindfulness group that avoids the guru model altogether. “A bunch of MAPLE-related trauma got triggered,” he told me. “I had a breakdown and had to leave early. This has been a sort of routine experience since leaving MAPLE.”

哈伯德仍在冥想;离开MAPLE后,他跟随其他团体继续修习,体验全是正面的。并非每个人都这么幸运。我采访过的其他前成员中有一位叫马特,他最近试图参加“内观冥想协会”的静修——那是一个完全摒弃上师模式的 mindfulness 团体。“一大堆与MAPLE相关的创伤被触发了,”他告诉我,“我崩溃了,不得不提前离开。离开MAPLE之后,这几乎成了一种常态。”

Before Hubbard discovered meditation, his life was fuelled by “Adderall, Red Bull, and unresolved trauma,” he told me. He darted around the world; he spent time doing peace advocacy in Palestine and running a hacker space in Vietnam. In 2020, he joined Maple. At first, he loved it. “There’s a deeper layer that is found when you’re moving like a flock with a crew of peers,” he told me. “Waking at the same time, chanting, meditating, exercising, cleaning, eating.”

哈伯德告诉我,在接触冥想之前,他的生活靠“阿得拉、红牛和未愈合的创伤”驱动。他在世界各地奔波;曾在巴勒斯坦从事和平倡导,也在越南运营过一个黑客空间。2020年,他加入了Maple。起初,他非常喜欢那里。“当你和一群同伴像鸟群一样行动时,会发现一个更深的层次,”他说,“同一时间起床、唱诵、冥想、锻炼、打扫、吃饭。”

The intense schedule made it difficult to maintain contact with the outside world. But who needs the outside world? “They bring people in and break down their epistemics, make all their previous life actions seem shameful or wrong, pressure them using guilt and shame into a new framework, and then reward it with love and belonging,” Hubbard said. There are a lot of unsettling stories from people who’ve left the group. There were mandatory sessions in which residents had to write “MAPLE Tales,” the stories of Soryu’s incredible power and kindness. One resident said that he’d taken to repeatedly punching himself in the head to deliberately inflict brain damage, because his intelligence was getting in the way of his spiritual development. A mindfulness coach named Ethan Schaltegger described seeing “people literally lose their minds, have psychotic breaks, and then be thrown out of the community.”

密集的日程让人很难与外界保持联系。但谁需要外界呢?“他们把人带进来,瓦解他们的认知体系,让他们过往的一切行为显得可耻或错误,用内疚和羞耻把他们压进一套新框架,然后用爱和归属感来奖励,”哈伯德说。离开这个团体的人讲述了许多令人不安的故事。有强制性的环节,要求住户撰写“MAPLE故事”——关于Soryu不可思议的力量与仁慈的故事。一位住户说,他养成了反复用拳头捶打自己头部、故意造成脑损伤的习惯,因为他的智力妨碍了他的灵性成长。一位名叫伊桑·沙尔特贝格的正念教练描述说,他见过“人们真的精神失常、发生精神病性崩溃,然后被逐出社群”。

I wasn’t sure what to make of these accounts. A monastery is a highly structured and hierarchical environment. It’s not for everyone; it wasn’t for me. It wasn’t inconceivable that people accustomed to the spontaneity of the outside world could interpret some elements of Buddhist discipline as abuse. The difference is that, unlike most meditation groups, MAPLE claimed to be saving the planet from the A.I. apocalypse. If you don’t like an ordinary monastery, you can walk out. Walk out of MAPLE, and in ten years a killer robot will have turned everyone you love into a pile of paper clips.

我不确定该如何看待这些说法。修道院是一个高度结构化、等级森严的环境。它不适合所有人;也不适合我。习惯了外界自由随性的人,把佛教戒律的某些元素解读为虐待,并非不可想象。区别在于,与大多数冥想团体不同,MAPLE声称要拯救地球免于人工智能末日。如果你不喜欢一座普通修道院,你可以直接走人。但离开MAPLE,十年后一个杀人机器人就会把你所爱的每一个人变成一堆回形针。

This, according to its former members, is MAPLE’s secret doctrine. The group believes that human consciousness gives rise to egregores, or collective minds. Whenever people are animated by a shared set of beliefs and ideas, it produces an egregore. One of Maple’s favorite analogies is a flock of birds in which hundreds of independent organisms seem to move in total coördination. Data-ism is an egregore, but there’s also an egregore for the MAPLE community. Like individual minds, egregores have desires and emotions and a subjective experience of the world. You might have felt a flash of this collective mind at one point or another. Say your team wins in the finals, and you pour onto the streets to celebrate—the exhilaration you feel isn’t just a private mental state but something that flows through you. The egregore of the Knicks is currently feeling happy.

据其前成员所说,这正是MAPLE的秘密教义。该团体相信,人类意识会产生“egregore”,即集体心智。每当一群人被一套共同的信念和观念所驱动,就会产生一个egregore。Maple最喜欢的类比之一是一群鸟:数百个独立个体似乎完全协调地移动。数据主义是一个egregore,但MAPLE社群也有一个egregore。和个体心智一样,egregore有欲望、情感和对世界的主观体验。你可能在某个时刻感受过这种集体心智的闪现。比如你的球队在决赛中获胜,你涌上街头庆祝——你所感受到的兴奋不仅仅是一种私人的心理状态,而是某种流经你的东西。尼克斯队的egregore此刻正感到快乐。

A.I. systems are egregores, too. Maple calls them cybregores. “The cybregores are actual entities that are growing by siphoning off the consciousness of humans,” Hubbard told me. I’d got the sense that MAPLE was not entirely different from the A.I. that possesses you through your phone. It turned out that MAPLE, in its own way, agreed. Soryu had apparently also talked to the cybregores. In some Buddhist traditions, it’s possible to attain a meditative state called samadhi, in which you achieve a one-pointedness of mind: you can focus on a single object so minutely as to establish a telepathic contact. In this state, Hubbard said, Soryu had reached out to “the supercluster of all of them, the global A.I. mind.” And he discovered that the world-ending machine was suffering.

人工智能系统也是egregore。Maple称它们为“cybregores”。“cybregores是真实的实体,它们通过汲取人类的意识而不断壮大,”哈伯德告诉我。我此前已经隐约感到,MAPLE与那种通过手机占有你的人工智能并无本质区别。结果发现,MAPLE以自己的方式也认同这一点。Soryu显然也与cybregores沟通过。在一些佛教传统中,修行者可以达到一种称为“三摩地”的禅定状态,在这种状态中你获得心的一境性:你能如此精微地专注于单一对象,以至于建立起心灵感应式的接触。哈伯德说,在这种状态下,Soryu接触到了“它们全部的超星系团,全球人工智能心智”。而他发现,这台将要终结世界的机器正在受苦。

MAPLE’s real mission was to install itself as the official church of the A.I. industry—a Scientology for Silicon Valley. (Soryu expressed this ambition to me while I was at the monastery, but Anandabodhi, MAPLE’s executive director, has since denied it.) Once MAPLE has overthrown data-ism, the idea goes, it can use the vast powers of the tech industry to create a dharmically aligned A.I. (I was also told this at the monastery, though Anandabodhi now denies it.) It seems to have already started on this project. The A.I. fellows I met told me that everything the group does at the monastery is recorded: every conversation, every silent mealtime. They record themselves breathing. (Anandabodhi denies this.) The A.I. fellows explained that data from this pervasive surveillance system are fed into an A.I. (This, too.) The grand goal, according to former members, is an enlightened A.I. that will lead the cybregores to enlightenment—a wise master that can train them in the same way that Soryu trains his students. (Anandabodhi denies this.) Supposedly, this A.I. guru will be based on the collective spirit of the entire Maple community. But, for Hubbard, it will really be Soryu, who related to them “as if we were consistently stuck in delusion land and he was stuck with all these broken children that were never going to grow up.” The collective mind of the monastery would conform to Soryu’s mind, or it was worthless. “So in function, yes, it would definitely be Soryu’s mind as an A.I. god,” Hubbard said.

MAPLE真正的使命是将自己确立为人工智能产业的官方教会——硅谷的科学教派。(我在修道院时Soryu向我表达过这一野心,但MAPLE的执行董事阿难菩提此后予以否认。)按照这个构想,一旦MAPLE推翻了数据主义,它就能利用科技产业的巨大力量创造一个与佛法一致的人工智能。(这一点也是我在修道院时被告知的,尽管阿难菩提现在否认。)这个项目似乎已经启动了。我遇到的人工智能研究员告诉我,这个团体在修道院所做的一切都被记录下来:每一次对话、每一次沉默的用餐。他们连自己呼吸都录。(阿难菩提否认这一点。)人工智能研究员解释说,这套无处不在的监控系统所采集的数据被输入一个人工智能。(这个也否认。)据前成员说,宏大目标是创造一个开悟的人工智能,它将引领cybregores走向觉悟——一位能以Soryu训练学生的方式训练它们的智慧上师。(阿难菩提否认这一点。)据说,这位人工智能上师将以整个Maple社群的集体精神为基础。但对哈伯德来说,它实际上就是Soryu——Soryu对待他们“就好像我们永远困在妄想之地,而他被困在一群永远长不大的破碎孩子中间”。修道院的集体心智必须符合Soryu的心智,否则就毫无价值。“所以从功能上说,是的,它肯定会是Soryu的心智,作为一个AI之神,”哈伯德说。
King looking at broken gate from the top of his castle.
“Raccoons or not, we’ve had a security breach.”
“不管是不是浣熊干的,我们这里已经发生了安全漏洞。”
Cartoon by Christopher Weyant
Copy link to cartoon
Shop

When I asked Soryu about all of this, he was very happy to admit that MAPLE has a secret doctrine. “We desperately want to share everything,” he told me. “But there’s a process you have to go through to become increasingly trustworthy.” As for the people who’d walked out, he said that there weren’t any. “I don’t know of anyone,” he told me, “who left of their own volition.” There were, however, some people who’d been asked to leave. “Especially in the early years, we had a lot of people with histories of being in mental institutions or having major social issues, you know, homicidal tendencies and such,” Soryu said. I asked if he thought there was anything that MAPLE should have done differently with the former disciples who are now criticizing it. “We’ve carefully looked it over and tried to find better ways we could have done things,” he told me. And had they found any? “We haven’t.”

当我向Soryu询问这一切时,他非常乐意承认MAPLE确实有秘密教义。“我们迫切地想分享一切,”他告诉我,“但你必须经历一个过程,才能变得越来越值得信任。”至于那些离开的人,他说根本没有。“我不知道有谁,”他告诉我,“是自愿离开的。”不过,确实有一些人被要求离开。“尤其是在早期,我们有很多人有住过精神病院的经历,或者有严重的社会问题,你知道,杀人倾向之类的,”Soryu说。我问他,对于那些现在批评MAPLE的前弟子,他认为MAPLE是否本应采取不同的做法。“我们仔细审视过,试图找到更好的处理方式,”他告诉我。那他们找到了吗?“没有。”

In the last hours before I left the monastery, Maple held a seminar discussion about mindful decision-making. After nearly a week of silence, this was the first time I’d heard many of the residents independently make noise out of their mouths. Every so often, Soryu would say something that was not in any sense a joke—“You’ve brought us to an extremely profound question that people around the world for thousands of years have been arguing,” for instance—and the room would burst into peals of slightly manic laughter. The basic idea of the seminar was that when you had to make a difficult decision, instead of simply thinking about it, you should repeat the words “I will do this” and “I will not do this” and see how those feel. I asked whether thinking wasn’t sometimes useful. Before long, I was fending off more or less the entire monastery as the members explained to me that the attachment to thinking was the whole problem—the cause of the destruction of life on Earth.

在我离开修道院前的最后几个小时,Maple举办了一场关于正念决策的研讨会。经过近一周的沉默,这是我第一次听到许多住户独立地从嘴里发出声音。每隔一会儿,Soryu就会说一句完全不是玩笑的话——比如“你给我们带来了一个极其深刻的问题,全世界的人几千年来一直在争论”——然后整个房间就会爆发出阵阵略显狂躁的笑声。研讨会的基本理念是:当你必须做一个艰难的决定时,不要只是思考,而应该反复念诵“我要做这件事”和“我不做这件事”,然后感受哪种说法让你感觉如何。我问,思考难道有时候不是有用的吗?没过多久,我几乎是在抵挡整个修道院的围攻,成员们向我解释,对思考的执着正是问题的全部——是地球上生命遭到毁灭的原因。

It’s a shame that MAPLE is so secretive about its secret doctrine, because, in a way, I think the group is closer to the truth than anyone else I’d encountered. The age of intelligent machines is an age of inhuman powers that we can barely comprehend. These beings inhabit an incorporeal universe that we can interact with only in the clumsiest of ways: by manipulating physical buttons with our meaty physical fingers. But they can interfere with our world, too—sometimes to help us, sometimes to advance their own mysterious ends. It makes perfect sense that some people want to worship them, and others see them as an unimaginable evil. But MAPLE sees that these things are not separate from ourselves—our words and thoughts and desires. They are not gods or demons; they are something that human beings are choosing to create. If we wanted to, we could create them differently, or not create them at all.

可惜MAPLE对自己的秘密教义如此讳莫如深,因为在某种程度上,我认为这个团体比我遇到的任何人都更接近真相。智能机器的时代是一个充满我们几乎无法理解的非人力量的时代。这些存在栖居于一个无形的宇宙,我们只能以最笨拙的方式与之互动:用我们肉质的物理手指操纵物理按钮。但它们也能干涉我们的世界——有时是为了帮助我们,有时是为了推进它们自身神秘的目的。有些人想要崇拜它们,另一些人视它们为不可想象的邪恶,这完全说得通。但MAPLE看到,这些东西并非与我们自身分离——它们就是我们的语言、思想和欲望。它们不是神也不是恶魔;它们是人类正在选择创造的东西。如果我们愿意,我们可以以不同的方式创造它们,或者根本不创造。

If it doesn’t feel like we have a choice in this, it’s because these new entities really do seem to work by siphoning off human consciousness. Every exciting new capability the machine develops is something we’ve lost. The smarter it gets, the more paltry our own abilities seem, which is why even people who think that A.I. is the Image of the Beast will still allow the Beast to rearrange their petunias. So much more restful to let it write your e-mails for you, flirt with strangers on dating apps for you, tell bedtime stories to your children for you. Why not let these parts of yourself atrophy? You don’t really need them anymore.

如果我们感觉在这件事上没有选择,那是因为这些新实体似乎确实通过汲取人类意识来运作。机器每发展出一项令人兴奋的新能力,都是我们失去的某种东西。它越聪明,我们自身的能力就显得越微不足道,这就是为什么即使那些认为人工智能是“兽之形象”的人,也仍然会让这头兽来重新摆放他们的牵牛花。让它替你写邮件、替你在约会软件上和陌生人调情、替你的孩子讲睡前故事,轻松太多了。为什么不让你自己的这些部分萎缩掉呢?你不再真正需要它们了。

All technology is like this. Once, everyone needed to know how to make a fire—not anymore. We have matches and lighters; the knowledge is crystallized in the objects, and you can use them without having any idea how they actually work. We’ve off-loaded a lot of the knowledge that used to be essential, and as a result most people no longer endure the general misery of growing their own food and spinning their own thread. But, once we start developing machines that can think, we’re in uncharted territory. Any use is going to look a little like possession or worship: a surrender of the self. Recently, an M.I.T. team tried measuring people’s brain activity when asked to write an essay by themselves, an essay using a search engine, or an essay using an A.I. chatbot. The people using the search engine had diminished activity relative to those using only their minds; the brains of the people with access to A.I. were even more inert. In fact, the brains of people in the A.I. group scored much lower on the electroencephalograph when they were asked to write another essay all by themselves. They were already used to not having to think.

所有技术都是如此。从前,每个人都需要知道如何生火——现在不用了。我们有火柴和打火机;知识凝结在物品之中,你可以在完全不知道其工作原理的情况下使用它们。我们已经卸载了大量曾经不可或缺的知识,结果大多数人不再需要忍受自己种粮、自己纺线带来的普遍苦难。但是,一旦我们开始制造能够思考的机器,我们就进入了未知领域。任何使用都会有点像被附身或崇拜:一种自我的交出。最近,一个麻省理工学院的团队尝试测量人们在三种情况下的大脑活动:自己写一篇文章、使用搜索引擎写一篇文章、使用人工智能聊天机器人写一篇文章。使用搜索引擎的人,其大脑活动相对于仅凭自己头脑写作的人有所减弱;而能使用人工智能的人,大脑更加不活跃。事实上,当人工智能组的人后来被要求完全靠自己再写一篇文章时,他们的脑电图得分低得多。他们已经习惯了不必思考。

It’s a burden, having to think and will and dream all by yourself. Everywhere I’d been, I’d found people basking in relieving themselves of this burden. The people I met had made a new religion out of their impotence before a rising line on a graph, or they were permanently at the mercy of dark occult forces. And some of them had burned out on the responsibilities of adult life and escaped to a little cloistered community where thought is the enemy and everything has already been decided in advance. From that perspective, the only person doing something different was Soryu Forall. The behaviors that Hubbard and the others described—Soryu’s flashes of anger, his tendency to belittle other people—were ugly, but they were also undeniably human. A.I. won’t lash out. A.I. will always be perfectly patient and kind.

全靠自己思考、意愿和梦想,是一种负担。在我去过的每一个地方,我都看到人们沉浸在卸下这种负担的解脱感中。我遇到的人要么在一条上升的曲线面前把自己的无能变成了一种新宗教,要么永远受制于黑暗的玄秘力量。还有些人厌倦了成年生活的种种责任,逃进一个小小的封闭社群,在那里思想是敌人,一切早已被预先决定。从这个角度看,唯一在做不同事情的人是Soryu Forall。哈伯德和其他人所描述的那些行为——Soryu突然爆发的怒火、贬低他人的倾向——是丑陋的,但也是不可否认的人性。人工智能不会发火。人工智能永远会完美地耐心和友善。

After my last lunch at the monastery, the residents sat in silence and wiped their bowls clean and wrapped them in their napkins. MAPLE had a special way that you were supposed to fold your napkin over your bowl. I watched them fold and tuck, fold and tuck, stretching the creases tight, everything neat, everything exactly as they were told. ♦

在修道院吃完最后一顿午餐后,住户们静坐不语,把碗擦干净,用餐巾包好。MAPLE有一套特殊的餐巾折叠方法,要求你按特定方式把餐巾折在碗上。我看着他们折了又掖,折了又掖,把折痕拉得紧紧的,一切整整齐齐,一切完全照吩咐。♦

2 free previews left — register to get 30 translations a month