人工智能的真正威脅

人工智能的真正威脅

經過葉夫根尼·莫羅佐夫 莫羅佐夫先生是《要拯救一切，點擊這裡：技術解決方案主義的愚蠢》一書的作者，也是即將推出的播客《聖地亞哥男孩》的 主持人。

2023 年 6 月 30 日 5 月，超過 350 名技術高管、研究人員和學者簽署了一份聲明，警告人工智能存在的危險。 簽署者警告說：“與流行病和核戰爭等其他社會規模的風險一樣，減輕人工智能帶來的滅絕 風險應該成為全球優先事項。” 在此之前，埃隆·馬斯克和蘋果聯合創始人史蒂夫·沃茲尼亞克等人簽署了另一封備受矚目的 信函，呼籲暫停先進人工智能係統的開發六個月。



與此同時，拜登政府敦促負責任的人工智能創新，並表示“為了抓住它提供的機遇”，我 們“必須首先管理它的風險”。在國會，參議員查克·舒默 (Chuck Schumer)呼籲就人工智能 的潛力和風險舉辦“同類首個”聽證會，這是來自行業高管、學者、民權活動人士和其他利 益相關者的速成課程。



人們對人工智能日益增長的焦慮並不是因為那些無聊但可靠的技術，這些技術可以自動完成我們的短信或指導機器人吸塵器躲避客廳中的障礙物。令專家們擔憂的是通用人工智能 （AGI）的崛起。 AGI 尚不存在，但一些人認為 OpenAI 的 ChatGPT 快速增長的功能表明它的出現已經不遠 了。



OpenAI 聯合創始人 Sam Altman將其描述為“通常比人類更聰明的系統”。構建這樣的 系統仍然是一項艱鉅的任務，有人說這是不可能的。但好處似乎確實很誘人。 想像一下，Roombas 不再只能用吸塵器吸地板，而是進化成多用途機器人，樂於在早上煮 咖啡或疊衣服，而無需編程來完成這些事情。 聽起來很吸引人。但如果這些 AGI Roomba 變得太強大，它們創造一塵不染的烏托邦的使 命可能會讓它們的塵土飛揚的人類主人變得混亂。至少我們跑得很好。 關於通用人工智能的討論中充斥著這樣的世界末日場景。



然而，由學者、投資者和企業家 組成的新興 AGI 遊說團體反駁說，一旦安全，AGI 將成為文明的福音。奧特曼先生是這場 競選活動的代言人，他開始了全球巡演以吸引立法者。今年早些時候，他寫道，通用人工 智能甚至可能會推動經濟發展，增加科學知識，並“通過增加豐富性來提升人類水平”。 這就是為什麼科技行業中有如此多的聰明人正在努力開發這項有爭議的技術，儘管如此令 人苦惱：不利用它來拯救世界似乎是不道德的。



他們受制於一種意識形態，認為這項新技術是不可避免的，而且在安全的情況下，是普遍有益的。它的支持者想不出更好的替代方案來修復人類並擴展其智慧。 但這種意識形態——稱之為 AGI 主義——是錯誤的。



AGI 的真正風險是政治性的，並且不會通過馴服叛逆的機器人來解決。最安全的 AGI 也無法提供其遊說團體所承諾的進步靈丹妙藥。通用人工智能主義將其出現描述為幾乎不可避免，它分散了人們對尋找增強智能的 更好方法的注意力。



AGI主義的支持者們並不知道，它只是一種更宏大意識形態的私生子，這種意識形態宣揚的 是，正如瑪格麗特·撒切爾 (Margaret Thatcher) 所說的那樣，除了市場之外別無選擇。 正如奧特曼暗示的那樣，通用人工智能——或者至少是急於建立它——更有可能為資本主義 最具破壞性的信條：新自由主義創造一個強大的（而且更時髦的）盟友，而不是像奧特曼先生暗示的那樣打破資本主義。 新自由主義的締造者著迷於私有化、競爭和自由貿易，希望通過市場和放鬆管制來激發和 改變停滯不前的、對勞動力友好的經濟。 其中一些轉變取得了成效，但也付出了巨大的代價。



多年來，新自由主義招致了許多批評者，他們將大衰退、金融危機、特朗普主義、英國脫歐等歸咎於它。 因此，拜登政府與這種意識形態保持距離也就不足為奇了，他們承認市場有時會出錯。基金會、智囊團和學術界甚至敢於想像後新自由主義的未來。 然而新自由主義還遠未消亡。更糟糕的是，它在通用人工智能主義中找到了盟友，而通用人工智能主義則強化並複制了它的主要偏見：私人行為者的表現優於公共行為者（市場偏見），適應現實勝過改變現實（適應偏見），效率勝過現實。社會問題（效率偏差）。 這些偏見完全顛覆了通用人工智能背後的誘人承諾：構建世界的追求不但不能拯救世界， 反而會讓事情變得更糟。方法如下。



AGI永遠無法克服市場對利潤的要求。 還記得優步以其低廉的價格吸引城市作為其公共交通系統嗎？ 一切都開始得很順利，優步承諾提供令人難以置信的廉價乘車服務，並提供自動駕駛汽車 和最低勞動力成本的未來。財力雄厚的投資者喜歡這個願景，甚至承擔了 Uber 數十億美元 的損失。 但當現實降臨時，自動駕駛汽車仍然是一個白日夢。投資者要求回報，Uber被迫提高價格。依靠它來取代公共巴士和火車的用戶被留在人行道上。 Uber 商業模式背後的新自由主義本能是私營部門可以比公共部門做得更好——這就是市場偏見。 不僅僅是城市和公共交通。醫院、警察部門甚至五角大樓越來越依賴矽谷來完成他們的任務。



有了通用人工智能，這種依賴只會加深，尤其是因為通用人工智能的範圍和野心是無限的。任何行政或政府服務都無法免受其破壞性承諾的影響。 此外，AGI 甚至不需要存在來吸引他們。無論如何，這是 Theranos 的教訓，這家初創公司 承諾通過革命性的血液檢測技術和前寵兒“解決”醫療保健問題美國的精英。它的受害者是 真實的，即使它的技術從來都不是。 在經歷了這麼多 Uber 和 Theranos 式的創傷之後，我們已經知道 AGI 的推出會帶來什麼。



它將包括兩個階段。一是高額補貼服務的魅力攻勢。然後是醜陋的緊縮，過度依賴的用戶和機構承擔了使他們盈利的成本。 一如既往，矽谷專家淡化了市場的作用。在最近一篇題為“為什麼人工智能將拯救世界”的 文章中，著名科技投資者馬克·安德森 (Marc Andreessen) 甚至宣稱人工智能“像任何其他技術一樣，由人擁有並由人控制”。 只有風險投資家才能使用如此精緻的委婉說法。大多數現代技術都屬於公司所有。他們 ——而不是神話中的“人民”——將成為通過拯救世界賺錢的人。 他們真的在拯救它嗎？到目前為止，記錄很差。



Airbnb 和 TaskRabbit 等公司被視為陷入困境的中產階級的救世主；特斯拉的電動汽車被視為解決地球變暖的良方。代餐奶昔 Soylent 的使命是“解決”全球飢餓問題，而 Facebook 則誓言要“解決”南半球的連接問題。 這些公司都沒有拯救世界。 十年前，我將這種現象稱為“解決方案主義”，但“數字新自由主義”也同樣合適。這種世界觀 根據營利性技術解決方案重新定義了社會問題。結果，屬於公共領域的擔憂被重新想像為市場中的創業機會。



AGI 主義重新點燃了這種解決主義的熱情。去年，奧特曼先生表示，“通用人工智能可能是人類生存所必需的”，因為“我們的問題似乎太大了”，我們“如果沒有更好的工具就無法解決”。他最近斷言通用人工智能將成為人類繁榮的催化劑。 但公司需要利潤，而這種仁慈，尤其是來自燒毀投資者數十億美元的無利可圖的公司的仁慈，並不常見。



OpenAI 已從微軟接受了數十億美元的投資，並考慮再籌集 1000 億美元來 建設 AGI。這些投資需要收回——以對抗該服務驚人的隱形成本。（2 月份的一項估計顯 示，ChatGPT 的運營費用為每天 700,000 美元。） 因此，醜陋的緊縮階段，以及為了使 AGI 服務盈利而大幅提價的階段，可能會在“豐富”和“繁榮”之前到來。但到那時，有多少公共機構會將變化無常的市場誤認為是負擔得起 的技術，並開始依賴 OpenAI 昂貴的產品呢？ 如果你不喜歡你的城鎮將公共交通外包給一家脆弱的初創企業，你是否希望它把福利服務、廢物管理和公共安全外包給可能更加不穩定的通用人工智能公司？



AGI 將減輕我們最棘手問題的痛苦，但不會解決它們。 新自由主義擅長利用技術來緩解社會的苦難。我記得 2017 年的一家創新科技企業承諾改善通勤者對芝加哥地鐵線路的使用。它提供獎勵來阻止地鐵乘客在高峰時段出行。它的創建者利用技術來影響需求方（乘客），認為供應方的結構性變化（例如籌集公共交通資金） 太困難了。科技將幫助芝加哥人適應城市日益惡化的基礎設施，而不是為了滿足公眾的需求而進行修復。 這就是適應偏差——希望藉助技術魔杖，我們可以對自己的困境變得麻木不仁。它是新自由主義對自力更生和韌性不懈鼓吹的產物。



信息很明確：做好準備，增強您的人力資本，並像初創企業一樣制定您的路線。AGI 主義 也呼應了這一點。比爾·蓋茨宣稱人工智能可以“幫助世界各地的人們改善生活”。 解決方案主義者的盛宴才剛剛開始：無論是對抗下一次流行病、孤獨症流行病還是通貨膨脹，人工智能已經被定位為釘住許多真實和想像的釘子的萬能錘子。然而，解決方案主義者的愚蠢行為所損失的十年揭示了此類技術修復的局限性。



可以肯定的是，矽谷的許多應用程序——用於監控我們的支出、卡路里和鍛煉製度——有時 會有所幫助。但他們大多忽視了貧困或肥胖的根本原因。如果不解決根源，我們就會陷入適應而非轉型的境地。 推動我們遵循步行習慣（有利於個人適應的解決方案）和理解為什麼我們的城鎮沒有公共空間可以行走（這是有利於集體和製度轉型的政治友好解決方案的先決條件）之間是有區別的。



但 AGI 主義與新自由主義一樣，認為公共機構缺乏想像力，而且生產力不高。他們應該適 應 AGI，至少按照 Altman 先生的說法，他最近表示，他對“我們的機構適應的速度”感到緊張——他補充說，這是“我們為什麼要開始部署這些技術”的部分原因。系統很早就開始了， 儘管它們確實很弱，這樣人們就有盡可能多的時間來做這件事。” 但機構只能適應嗎？他們不能製定自己的變革議程來提高人類的智力嗎？或者我們使用機構只是為了減輕矽谷自身技術的風險嗎？



AGI 破壞了公民道德並放大了我們已經不喜歡的趨勢。 對新自由主義的一個常見批評是，它使我們的政治生活變得扁平化，圍繞效率進行重新安排。1960 年的一篇文章《社會成本問題》已成為新自由主義經典，其中宣揚污染工廠及其 受害者不應費心將糾紛告上法庭。這樣的鬥爭是低效的——到底誰需要正義呢？——並阻礙 市場活動。相反，雙方應該私下討價還價並繼續開展業務。 這種對效率的執著正是我們通過讓最嚴重的罪犯繼續像以前一樣來“解決”氣候變化的方 式。避免監管束縛的方法是製定一項計劃（在本例中是對碳徵稅），讓污染者購買信用額 度以匹配他們排放的額外碳。 這種效率文化，其中市場衡量事物的價值並取代正義，不可避免地腐蝕公民美德。



由此產生的問題隨處可見。學者們擔心，在新自由主義下，研究和教學已經成為商品。醫生們感嘆，醫院優先考慮更有利可圖的服務，例如擇期手術，而不是緊急護理。記者們討厭他們的文章的價值是用眼球來衡量的。 現在想像一下，向這些受人尊敬的機構——大學、醫院、報紙——釋放通用人工智能，並肩 負著“修復”它們的崇高使命。他們隱含的公民使命對 AGI 來說仍然是不可見的，因為這些使命即使在他們的年度報告中也很少被量化——這些材料用於培訓 AGI 背後的模型 畢竟，誰喜歡吹噓他的文藝復興歷史課只有少數學生呢？或者她關於某個遙遠國家的腐敗問題的文章只有十幾個頁面瀏覽量？效率低下且無利可圖，即使在當前系統中，這些異常值也奇蹟般地生存了下來。



該機構的其他部門悄悄地補貼他們，優先考慮利潤驅動的“效率”之外的價值觀。 AGI 烏托邦中的情況還會如此嗎？或者通過通用人工智能來修復我們的機構是否會像將它們交給無情的顧問一樣？他們也提供數據支持的“解決方案”以最大限度地提高效率。但這些解決方案往往無法抓住機構核心的價值觀、使命和傳統之間混亂的相互作用——如果你 只觸及數據的表面，這種相互作用是很難被發現的。



事實上，類 ChatGPT 服務的卓越性能在設計上就是拒絕在數據表面之外的更深層次上把握現實。因此，儘管早期的人工智能係統依賴於明確的規則，並需要像牛頓這樣的人對重力進行理論化——詢問蘋果如何以及為何下落，但像 AGI 這樣的新系統只是通過觀察數百萬個蘋果下落到地面來學習預測重力的影響。 然而，如果 AGI 看到的只是資金短缺的機構為生存而戰，它可能永遠無法推斷出它們的真實精神。祝你好運，通過觀察已變成利潤中心的醫院來辨別希波克拉底誓言的含義。



瑪格麗特·撒切爾的另一句著名的新自由主義格言是“不存在社會這樣的東西”。 AGI 遊說團體無意中也認同這一悲觀觀點。對他們來說，值得複制的智力是個人頭腦中發生的事情的函數，而不是整個社會的函數。 但人類智力不僅是基因和個人才能的產物，也是政策和製度的產物。在國會圖書館獲得獎學金比在沒有書店甚至沒有像樣的 Wi-Fi 的地方做幾份工作要容易得多。 更多的獎學金和公共圖書館將為提高人類智力創造奇蹟，這似乎並沒有那麼有爭議。但對 於矽谷的解決方案論者來說，增強智能主要是一個技術問題——因此 AGI 令人興奮



然而，如果通用人工智能主義確實是其他意義上的新自由主義，那麼我們應該準備好看到 更少——而不是更多——支持情報的機構。畢竟，他們是那個可怕的“社會”的殘餘，對於新自由主義者來說，這個社會並不真正存在。AGI 增強智能的宏偉計劃最終可能會縮小它。 由於這種解決方案主義偏見，即使是圍繞通用人工智能的看似創新的政策想法也無法令人 興奮。以最近提出的“人工智能安全曼哈頓計劃”為例。這是基於 AGI 之外別無選擇的錯誤 想法 但是，如果政府資助曼哈頓文化和教育項目以及培育它們的機構，我們對增強智力的追求 難道不會更有效嗎？



如果沒有這樣的努力，我們現有公共機構的大量文化資源就有可能成為通用人工智能初創 企業的培訓數據集，從而強化社會不存在的謊言。 根據機器人叛亂如何（以及是否）展開，AGI 可能會或可能不會證明存在的威脅。但由於其反社會傾向和新自由主義偏見，通用人工智能主義已經是這樣的：我們不需要等待神奇的 Roomba 來質疑它的原則。





The True Threat of Artificial Intelligence https://www.nytimes.com/2023/06/30/opinion/artificial-intelligencedanger.html 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 2/7 Discussions of A.G.I. are rife with such apocalyptic scenarios. Yet a nascent A.G.I. lobby of academics, investors and entrepreneurs counter that, once made safe, A.G.I. would be a boon to civilization. Mr. Altman, the face of this campaign, embarked on a global tour to charm lawmakers. Earlier this year he wrote that A.G.I. might even turbocharge the economy, boost scientific knowledge and “elevate humanity by increasing abundance.” This is why, for all the hand-wringing, so many smart people in the tech industry are toiling to build this controversial technology: not using it to save the world seems immoral. They are beholden to an ideology that views this new technology as inevitable and, in a safe version, as universally beneficial. Its proponents can think of no better alternatives for fixing humanity and expanding its intelligence. But this ideology — call it A.G.I.-ism — is mistaken. The real risks of A.G.I. are political and won’t be fixed by taming rebellious robots. The safest of A.G.I.s would not deliver the progressive panacea promised by its lobby. And in presenting its emergence as all but inevitable, A.G.I.-ism distracts from finding better ways to augment intelligence. Unbeknown to its proponents, A.G.I.-ism is just a bastard child of a much grander ideology, one preaching that, as Margaret Thatcher memorably put it, there is no alternative, not to the market. Rather than breaking capitalism, as Mr. Altman has hinted it could do, A.G.I. — or at least the rush to build it — is more likely to create a powerful (and much hipper) ally for capitalism’s most destructive creed: neoliberalism. Fascinated with privatization, competition and free trade, the architects of neoliberalism wanted to dynamize and transform a stagnant and labor-friendly economy through markets and deregulation. Some of these transformations worked, but they came at an immense cost. Over the years, neoliberalism drew many, many critics, who blamed it for the Great Recession and financial crisis, Trumpism, Brexit and much else. It is not surprising, then, that the Biden administration has distanced itself from the ideology, acknowledging that markets sometimes get it wrong. Foundations, think tanks and academics have even dared to imagine a post-neoliberal future. Yet neoliberalism is far from dead. Worse, it has found an ally in A.G.I.-ism, which stands to reinforce and replicate its main biases: that private actors outperform public ones (the market bias), that adapting to reality beats transforming it (the adaptation bias) and that efficiency trumps social concerns (the efficiency bias). These biases turn the alluring promise behind A.G.I. on its head: Instead of saving the world, the quest to build it will make things only worse. Here is how. 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 3/7 A.G.I. will never overcome the market’s demands for profit. Remember when Uber, with its cheap rates, was courting cities to serve as their public transportation systems? It all began nicely, with Uber promising implausibly cheap rides, courtesy of a future with self-driving cars and minimal labor costs. Deep-pocketed investors loved this vision, even absorbing Uber’s multibillion-dollar losses. But when reality descended, the self-driving cars were still a pipe dream. The investors demanded returns and Uber was forced to raise prices. Users that relied on it to replace public buses and trains were left on the sidewalk. The neoliberal instinct behind Uber’s business model is that the private sector can do better than the public sector — the market bias. It’s not just cities and public transit. Hospitals, police departments and even the Pentagon increasingly rely on Silicon Valley to accomplish their missions. With A.G.I., this reliance will only deepen, not least because A.G.I. is unbounded in its scope and ambition. No administrative or government services would be immune to its promise of disruption. Moreover, A.G.I. doesn’t even have to exist to lure them in. This, at any rate, is the lesson of Theranos, a start-up that promised to “solve” health care through a revolutionary blood-testing technology and a former darling of America’s elites. Its victims are real, even if its technology never was. After so many Uber- and Theranos-like traumas, we already know what to expect of an A.G.I. rollout. It will consist of two phases. First, the charm offensive of heavily subsidized services. Then the ugly retrenchment, with the overdependent users and agencies shouldering the costs of making them profitable. As always, Silicon Valley mavens play down the market’s role. In a recent essay titled “Why A.I. Will Save the World,” Marc Andreessen, a prominent tech investor, even proclaims that A.I. “is owned by people and controlled by people, like any other technology.” Only a venture capitalist can traffic in such exquisite euphemisms. Most modern technologies are owned by corporations. And they — not the mythical “people” — will be the ones that will monetize saving the world. And are they really saving it? The record, so far, is poor. Companies like Airbnb and TaskRabbit were welcomed as saviors for the beleaguered middle class; Tesla’s electric cars were seen as a remedy to a warming planet. Soylent, the meal-replacement shake, 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 4/7 embarked on a mission to “solve” global hunger, while Facebook vowed to “solve” connectivity issues in the Global South. None of these companies saved the world. A decade ago, I called this solutionism, but “digital neoliberalism” would be just as fitting. This worldview reframes social problems in light of for-profit technological solutions. As a result, concerns that belong in the public domain are reimagined as entrepreneurial opportunities in the marketplace. A.G.I.-ism has rekindled this solutionist fervor. Last year, Mr. Altman stated that “A.G.I. is probably necessary for humanity to survive” because “our problems seem too big” for us to “solve without better tools.” He’s recently asserted that A.G.I. will be a catalyst for human flourishing. But companies need profits, and such benevolence, especially from unprofitable firms burning investors’ billions, is uncommon. OpenAI, having accepted billions from Microsoft, has contemplated raising another $100 billion to build A.G.I. Those investments will need to be earned back — against the service’s staggering invisible costs. (One estimate from February put the expense of operating ChatGPT at $700,000 per day.) Thus, the ugly retrenchment phase, with aggressive price hikes to make an A.G.I. service profitable, might arrive before “abundance” and “flourishing.” But how many public institutions would mistake fickle markets for affordable technologies and become dependent on OpenAI’s expensive offerings by then? And if you dislike your town outsourcing public transportation to a fragile start-up, would you want it farming out welfare services, waste management and public safety to the possibly even more volatile A.G.I. firms? A.G.I. will dull the pain of our thorniest problems without fixing them. Neoliberalism has a knack for mobilizing technology to make society’s miseries bearable. I recall an innovative tech venture from 2017 that promised to improve commuters’ use of a Chicago subway line. It offered rewards to discourage metro riders from traveling at peak times. Its creators leveraged technology to influence the demand side (the riders), seeing structural changes to the supply side (like raising public transport funding) as too difficult. Tech would help make Chicagoans adapt to the city’s deteriorating infrastructure rather than fixing it in order to meet the public’s needs. This is the adaptation bias — the aspiration that, with a technological wand, we can become desensitized to our plight. It’s the product of neoliberalism’s relentless cheerleading for self-reliance and resilience. 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 5/7 The message is clear: gear up, enhance your human capital and chart your course like a start-up. And A.G.I.-ism echoes this tune. Bill Gates has trumpeted that A.I. can “help people everywhere improve their lives.” The solutionist feast is only getting started: Whether it’s fighting the next pandemic, the loneliness epidemic or inflation, A.I. is already pitched as an all-purpose hammer for many real and imaginary nails. However, the decade lost to the solutionist folly reveals the limits of such technological fixes. To be sure, Silicon Valley’s many apps — to monitor our spending, calories and workout regimes — are occasionally helpful. But they mostly ignore the underlying causes of poverty or obesity. And without tackling the causes, we remain stuck in the realm of adaptation, not transformation. There’s a difference between nudging us to follow our walking routines — a solution that favors individual adaptation — and understanding why our towns have no public spaces to walk on — a prerequisite for a politics-friendly solution that favors collective and institutional transformation. But A.G.I.-ism, like neoliberalism, sees public institutions as unimaginative and not particularly productive. They should just adapt to A.G.I., at least according to Mr. Altman, who recently said he was nervous about “the speed with which our institutions can adapt” — part of the reason, he added, “of why we want to start deploying these systems really early, while they’re really weak, so that people have as much time as possible to do this.” But should institutions only adapt? Can’t they develop their own transformative agendas for improving humanity’s intelligence? Or do we use institutions only to mitigate the risks of Silicon Valley’s own technologies? A.G.I. undermines civic virtues and amplifies trends we already dislike. A common criticism of neoliberalism is that it has flattened our political life, rearranging it around efficiency. “The Problem of Social Cost,” a 1960 article that has become a classic of the neoliberal canon, preaches that a polluting factory and its victims should not bother bringing their disputes to court. Such fights are inefficient — who needs justice, anyway? — and stand in the way of market activity. Instead, the parties should privately bargain over compensation and get on with their business. This fixation on efficiency is how we arrived at “solving” climate change by letting the worst offenders continue as before. The way to avoid the shackles of regulation is to devise a scheme — in this case, taxing carbon — that lets polluters buy credits to match the extra carbon they emit. 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 6/7 This culture of efficiency, in which markets measure the worth of things and substitute for justice, inevitably corrodes civic virtues. And the problems this creates are visible everywhere. Academics fret that, under neoliberalism, research and teaching have become commodities. Doctors lament that hospitals prioritize more profitable services such as elective surgery over emergency care. Journalists hate that the worth of their articles is measured in eyeballs. Now imagine unleashing A.G.I. on these esteemed institutions — the university, the hospital, the newspaper — with the noble mission of “fixing” them. Their implicit civic missions would remain invisible to A.G.I., for those missions are rarely quantified even in their annual reports — the sort of materials that go into training the models behind A.G.I. After all, who likes to boast that his class on Renaissance history got only a handful of students? Or that her article on corruption in some faraway land got only a dozen page views? Inefficient and unprofitable, such outliers miraculously survive even in the current system. The rest of the institution quietly subsidizes them, prioritizing values other than profit-driven “efficiency.” Will this still be the case in the A.G.I. utopia? Or will fixing our institutions through A.G.I. be like handing them over to ruthless consultants? They, too, offer data-bolstered “solutions” for maximizing efficiency. But these solutions often fail to grasp the messy interplay of values, missions and traditions at the heart of institutions — an interplay that is rarely visible if you only scratch their data surface. In fact, the remarkable performance of ChatGPT-like services is, by design, a refusal to grasp reality at a deeper level, beyond the data’s surface. So whereas earlier A.I. systems relied on explicit rules and required someone like Newton to theorize gravity — to ask how and why apples fall — newer systems like A.G.I. simply learn to predict gravity’s effects by observing millions of apples fall to the ground. However, if all that A.G.I. sees are cash-strapped institutions fighting for survival, it may never infer their true ethos. Good luck discerning the meaning of the Hippocratic oath by observing hospitals that have been turned into profit centers. Margaret Thatcher’s other famous neoliberal dictum was that “there is no such thing as society.” The A.G.I. lobby unwittingly shares this grim view. For them, the kind of intelligence worth replicating is a function of what happens in individuals’ heads rather than in society at large. But human intelligence is as much a product of policies and institutions as it is of genes and individual aptitudes. It’s easier to be smart on a fellowship in the Library of Congress than while working several jobs in a place without a bookstore or even decent Wi-Fi. 2023/7/3 下午4:32 Opinion | The True Threat of Artificial Intelligence - The New York Times https://www.nytimes.com/2023/06/30/opinion/artificial-intelligence-danger.html 7/7 It doesn’t seem all that controversial to suggest that more scholarships and public libraries will do wonders for boosting human intelligence. But for the solutionist crowd in Silicon Valley, augmenting intelligence is primarily a technological problem — hence the excitement about A.G.I. However, if A.G.I.-ism really is neoliberalism by other means, then we should be ready to see fewer — not more — intelligence-enabling institutions. After all, they are the remnants of that dreaded “society” that, for neoliberals, doesn’t really exist. A.G.I.’s grand project of amplifying intelligence may end up shrinking it. Because of such solutionist bias, even seemingly innovative policy ideas around A.G.I. fail to excite. Take the recent proposal for a “Manhattan Project for A.I. Safety.” This is premised on the false idea that there’s no alternative to A.G.I. But wouldn’t our quest for augmenting intelligence be far more effective if the government funded a Manhattan Project for culture and education and the institutions that nurture them instead? Without such efforts, the vast cultural resources of our existing public institutions risk becoming mere training data sets for A.G.I. start-ups, reinforcing the falsehood that society doesn’t exist. Depending on how (and if) the robot rebellion unfolds, A.G.I. may or may not prove an existential threat. But with its antisocial bent and its neoliberal biases, A.G.I.-ism already is: We don’t need to wait for the magic Roombas to question its tenets. Evgeny Morozov, the author of “To Save Everything, Click Here: The Folly of Technological Solutionism,” is the founder and publisher of The Syllabus and the host of the podcast “The Santiago Boys.” The Times is committed to publishing a diversity of letters to the editor. We’d like to hear what you think about this or any of our articles. Here are some tips. And here’s our email: letters@nytimes.com. Follow The New York Times Opinion section on Facebook, Twitter (@NYTopinion) and Instagram. A version of this article appears in print on , Section SR, Page 6 of the New York edition wit