这一警告直白而明确：“还没有人愿意承认计算机是人类正式的一员，但面对现实吧——它们正越来越接近。”那是在1961年。“人工智能”一词当时尚未普遍使用，但现已停刊的下午报《洛杉矶镜报》当时向读者报道了南加州一家计算机制造公司一项令人费解的新进展。“信不信由你，”报道写道，“在格伦代尔的通用精密仪器公司利布拉斯科普分部，那些负责照料和喂养这些电子怪物的人，竟然让这些该死的家伙写起了诗。”
The warning was stark and unambiguous: “No one is quite ready to admit computing machines as full-fledged members of the human race but let’s face it—they’re getting closer.”That was in 1961. The term “artificial intelligence” wasn’t in common use, but the Los Angeles Mirror, a now-defunct afternoon newspaper, was telling readers about a puzzling development at a Southern California computer manufacturing company.“Believe it or not,” the story said, “the men who handle the care and feeding of the electronic monsters at the Librascope division of General Precision, Inc., in Glendale have got the darn things writing poetry.”
报道称，这些诗或许无法媲美罗伯特·弗罗斯特或卡尔·桑德堡，但这不是重点：“这股写诗热潮是一个研究项目，旨在寻找更好、更快、更便宜的方法，来激发机器做出智能反应……计算机能储存比人更多的信息，且能完整回忆，能关联更多数据，能比人比较更多数量的可能答案。”其含义很清楚：如果研究人员充分开发计算机的能力，它可能在思考、规划和行动方面超越人类。
The poems might not emulate Robert Frost or Carl Sandburg, the story said, but that wasn’t the point: “The poetry kick is a research project directed toward finding better, faster and less expensive ways of stimulating the machines into intelligent reactions. . . . The computer can store more information, with total recall, than a human, can correlate more data, and compare a much larger number of possible answers than a person.”The implications were clear: If researchers developed the computer’s capabilities fully enough, it might surpass humans in thinking, planning and acting.
1961年末，大多数美国人关注的是其他事情——载人太空竞赛、彩色电视的问世、持续整个夏天的米奇·曼特尔与罗杰·马里斯本垒打追逐战的余温——因此加州那间计算机实验室里的动向大多被视为一件奇事。但《镜报》提醒说，可以想象，有朝一日“这台机器可能会提出新颖、灵活的想法”——也就是说，能够自行思考。而作为对当今人们担忧AI智能体违抗人类主人这一情形的诡异先声，用于该项目的首台机器“超出了它的电子裤子”——变得如此聪明，以至于需要更大的硬件。这台机器的部分输出被当作消遣娱乐来对待。
Most Americans were paying attention to other things in late 1961—the race to send men to space, the arrival of color television, the afterglow of the summerlong Mickey Mantle-Roger Maris home-run chase—so the goings-on at the computer lab in California were mostly regarded as a curiosity. But it was conceivable, the Mirror cautioned, that someday “the machine is likely to come up with fresh, flexible ideas”—that is, to think on its own. And in an eerie precursor to today’s concerns about AI agents defying their human masters, the first machine used for the project “outgrew its electronic britches”—became so smart that it required larger hardware. Some of the machine’s output was treated as an amusement.
其中一首早期诗作写道：“哦，别为这温顺的汁液惊慌／最后，我的几件夹克确实不信任那只鹅。”但《镜报》预测：“假以时日，或许计算机会写出一本书。”几天后，本迪克斯公司的高管汤姆·霍兰告知该报，他公司的G-15计算机已经做到了这一点：一本511页的精装书，名为《角度分度表》，内容是关于“将圆分成任意数量的等份”。
One of its early poems read: “Oh, panic not to this docile juice / Finally, few of my jackets did distrust the goose.”But the Mirror predicted: “In time, perhaps a computer will write a book.”A few days later Tom Horan, an executive of Bendix Corp., let the paper know that his company’s G-15 computer had already done just that: a 511-page hardcover volume called “Angular Indexing Tables” about “dividing a circle into any number of equal divisions.”
他说，两名员工向计算机输入了基本信息，机器在112小时内写完了这本书。霍兰说，由于其晦涩的主题，“这本书永远不会登上畅销书榜”，但它已经售出300多本。如果机器有朝一日发展出足够的智能——尽管可能是人工的——使我们变得多余，那么纯粹血肉之躯的男男女女还有任何希望吗？《镜报》资深专栏作家马特·温斯托克带来了一线乐观：“我感觉，如果计算机有朝一日被接纳为人类的一员，它们会想退出。”格林先生的著作包括《雪佛兰之夏，冰雪皇后之夜》。
Two employees, he said, had fed the computer basic information, and the machine had written the book in 112 hours. Horan said that because of its arcane subject matter “the book will never make the bestseller list,” but it had already sold more than 300 copies. If the machines were someday to develop enough intelligence, artificial though it might be, to render us obsolete, was there any hope for mere flesh-and-blood men and women? Matt Weinstock, a longtime Mirror columnist, offered a glimmer of optimism: “I get the feeling that if and when computers are admitted to the human race they’ll want out.”Mr. Greene’s books include “Chevrolet Summers, Dairy Queen Nights.”