AI寒冬史-History of the first AI Winter

AI has a long history. One can argue it even started long before the term was first coined; mostly in stories and later in actual mechanical devices called automata. This chapter only covers events relevant to the periods of AI winters without being too exhaustive in hope to extract knowledge that can be applied today.

人工智能有着很长的历史。有些人甚至说，在AI这个术语发明之前人工智能就已经存在了。尽管那时它只是存在于故事和后来一种被称之为自动机的机械装置当中。本文主要涵盖了AI寒冬的相关事件；它没有特别详尽，你也别指望能从中提取一些知识，以便在当下使用。

Events leading to the first AI Winter

第一次AI寒冬的导火索

To aid understanding the phenomenon of AI Winters, the events leading up to them are examined.

为了帮助理解AI寒冬的相关现象，我们会逐一检查导致它们的事件。

Beginnings of the AI Field in the 1950s

人工智能开端于20世纪50年代

Many early ideas about thinking machines appeared in the late 1940s to ’50s by people like Turing or Von Neumann. Turing tried to frame the questions of “Can machines think?” differently and created the imitation game, now famously called the Turing Test.

关于思维机器的很多早期想法出现于20世纪40年代末至50年代这段时间内，像图灵或者冯诺依曼等人就提出了许多相关理论。图灵试图使用不同的方式定义“机器能思考吗”这一问题，并提出了“模仿游戏”，也就是现在被称为“图灵测试”的假说。

In 1955, Arthur Samuel wrote a program that could play checkers very well. A year later, it even appeared on television. It used a combination of a tree search with heuristics and learned weights. Samuel handcrafted the heuristics inspired by a book from checkers experts. He used a learning algorithm he called “temporal-difference learning” where the weights are adjusted using the “error” between the score initially calculated and the score after the search was completed.

1955年，亚瑟塞缪尔编写了一个跳棋玩得很棒的程序。一年之后，它甚至出现在了电视上。它的算法包括了启发式的树搜索和学习权重组合等方法。其中，启发式的树搜索这一方法的灵感来自跳棋专家的一本著作。另外，他还使用了一种被他称为“差分学习”的学习算法；在此算法中，通过最初估算的分数和最终搜索完成后得到的分数之间的“误差”进行计算，从而完成对权重的调整。

In 1954, one of the first experiments in machine translation was executed. It used a 250-word dictionary for translation combined with syntactical ysis. Translations between English and Russian were demonstrated. The New York Times commented

1954年，研究人员进行了机器翻译领域的第一次试验。结合语法的同时，算法使用了一个250词的词典进行翻译。算法演示了英语和俄语之间的翻译工作。对此评论说：

“This admittedly will amount to a crude word-for-word translation … but will nevertheless be extremely valuable, the designers say, for such purposes as scientists translations of foreign technical papers in which vocabulary is far more of a problem than syntax .”

“这无疑是一个逐字进行的粗糙翻译……但研究人员说，它仍然具有很高的价值。对科学家们而言，在翻译外国技术文献时，其中的词汇问题要比语法问题重要的多。”

By then, the designers thought that most of the work was done with only some all errors to fix. Hutchin noted that it was the most far-reaching coverage that machine translation has ever received. For this reason, it generated tremendous hype and made it easier to obtain funding for the following work.

在当时，研究人员认为大部分的工作已经完成，他们只需要进行一些小修小补的工作即可。Hutchin指出，这是机器翻译所获得的最为广泛的报道。由于这一原因，它引发了巨大的炒作，并且使得这一领域在随后更容易获得资金支持。

AI research gained much funding from U.S. Defense Establishments (ONR and ARPA, later called DARPA) in the hope that these technologies would be useful for the U.S. Navy. At the time, there was a substantial amount of enthusia and optimi regarding the state of AI. The field of machine translation was especially important during the cold war, as the government had a big interest in automatic translation from Russian to English.

人工智能研究从美国国防机构（ONR和ARPA，后来被称为DARPA）那里获得了大量资金，他们希望这些技术能够帮助到美国海军。当时，所有人对人工智能的发展状态都是热情而乐观的。在冷战期间，机器翻译这一领域显得尤为重要，因为美国希望能够实现从俄语到英语的自动翻译。

Early experiments served as an inspiration to create the Dartmouth Summer Project in 1956, where the term AI was coined. The summer project was held under the motif that “Every aspect of learning or any other feature of intelligence can in principle be so precisely described that a machine can be made to simulate it”. Researchers from many different fields were invited and a lot of different ideas, papers, and concepts were put forward. Though progress was made some were disappointed, for instance, McCarthy said in response to the workshop “[the] main reason the Workshop did not live up to my expectations is that AI is harder than we thought.”

早期的一些试验促进了1956年的达特茅斯夏季项目，正是在此项目中，AI这个术语被发明了出来。这一夏季项目的主题是，“学习的每个方面或者智能的所有特征，原则上都可以被精确地描述，从而使得机器可以进行模仿”。项目邀请了许多不同领域的研究人员，大量不同的想法、论文和概念在项目中被提出。虽然项目取得了很多进展，但是有些人仍然感到失望。例如，麦卡锡在研讨会上曾表示，“研讨会没有让我感到满意的主要原因是，人工智能比我们想象的要难得多。”

In 1957, Rosenblatt invented perceptrons, a type of neural network where binary neural units are connected via adjustable weights. He was inspired by the work of neuroscience in the 1940s, which led him to create a crude replication of the neurons in the brain.

1957年，罗森布拉特发明了感知器。它是神经网络的一种形式，其中的（二元）神经元之间的连接权重是可调节的。在受到20世纪40年代神经科学相关工作的启发之后，他对大脑神经元（的连接方式）进行粗略，从而创造了这一模型。

He tried many different layouts and learning algorithms. One type of perceptron was called series-coupled, which would in today terms refer to the standard feedforward layout of a neural network where data flows from input to output. A prominent layout was what he called alpha perceptron, a three-layer series-coupled network where the three layers, in this case, included the input and output layer. Computers at the time would have been too slow to run the perceptron, so Rosenblatt built a special purpose machine with adjustable resistors (potentiometers) controlled by little motors. The apparatus was able to learn to classify different images of shapes or letters. The New York Times reported on the perceptron

他尝试了许多不同的连接方式和学习算法。其中的一种感知器被称为串联耦合类型，这一类型在今天的术语中被称为标准前馈神经网络，即数据从输入流向输出。在这一类型中，有一种被他称为alpha感知器的网络格外突出。这是一种三层串联耦合的网络，其中包括了输入层和输出层。由于当时的计算机运行感知器的速度太慢，罗森布拉特建造了一台由小型电机驱动，带有可调电阻器（电位器）的专用机器。该装置经过学习后能够对不同形状、字母的图片进行分类。对此进行了报道。

“The Navy revealed the embryo of an electronic computer that it expects will be able to walk, talk, see, write, reproduce itself and be conscious of its existence.”

“海军透露消息说，他们研发了一种新型电子计算机的雏形。他们预期该电子计算机将来能够走路、说话、观察、书写和，并能够意识到自我的存在。”

In the same year (1957), Simon summarized the current progress on AI like this

在同一年（1957年），西蒙对人工智能的最新进展进行总结说：

“It is not my aim to surprise or shock you—but the st way I can summarize is to say that there are now in the world machines that think, that learn, and that create. Moreover, their ability to do these things is going to increase rapidly until—in a visible future—the range of problems they can handle will be coextensive with the range to which the human mind has been applied.”

“我的目的不是要让你感到吃惊或者震惊，但对此（人工智能）进行最简单的总结就是，在这个世界上，现在已经存在可以进行思考、学习和创造的机器了。此外，它们做事的能力正在迅速增加，在可见的未来内，它们将能够处理人类所能处理的大部分问题。”

The Quiet Decade

寂静的十年

After the increases in funding and enthusia for machine translation in the 1950s and early 1960s, progress

stalled. Hutchins called the period of 1967 to 1976 the quiet decade of machine translation. Bar-Hillel said that machine translation was not feasible. He demonstrated that computers would need too much information about the world for correct translation, which he thought was “utterly chimerical and hardly deserves any further discussion.” The Automatic Language Processing Advisory Committee concluded in 1964 that there was no immediate or predictable prospect of useful machine translation. A report in 1966 concluded that “there has been no machine translation of general scientific text, and none is in immediate prospect,” which led to a cut in funding for all academic translation projects.

在20世纪50年代和60年代的早期，大量资金和支持被投入到机器翻译的研究中，但是进展寥寥。哈钦斯将1967年至1976年称为机器翻译“寂静的十年”。Bar-Hillel认为机器翻译完全是不可行的。他证明，计算机在能够正确进行翻译之前需要海量的关于这个世界的正确信息，因此他认为“（机器翻译）完全就是空想，没有进一步进行讨论的价值”。自动语言处理委员会在1964年得出结论，（可用的）机器翻译的前景，当下不存在，未来也不可预测。1966年的一份报告总结说，“机器翻译仍然无法处理一般的科学文本，其前景尚不明朗”。这导致所有学术机器翻译项目的资金都被大幅削减。

The disappointments in machine translation were especially detrimental to the field of AI because the U.S. Defense Establishment hoped for usable systems to emerge, but similar patterns were noticed in other fields of AI too. In 1965, Dreyfus paralleled AI with alchemy. He stud the achievements of several fields at the time and concluded “An overall pattern is taking shape an early, dramatic success based on the easy performance of tasks, or low-quality work on complex tasks, and then diminishing returns, disenchantment, and, in some cases, pessimi.”

机器翻译的失败对人工智能领域产生了非常严重的不良影响，因为美国的国防机构（进行投资的目的是）本来希望能够产出可用的机器翻译系统；而在人工智能的其他领域，类似的事情也在上演。1965年，德雷福斯将人工智能称为炼金术。他研究了几个不同领域取得的成就之后总结说，“整体来看（这些研究的）模式是这样的：这些早期的戏剧性的成功，要么基于简单任务的轻巧表现，要么基于复杂任务的低质量表现。所以投资的收益递减，失望，以及在某些情况下的悲观情绪，都是不可避免的。”

The quiet decade of machine translation was one of the most significant events that gave rise to the first AI winter. Another was the fall of the connectionist movement.

机器翻译“寂静的十年”是引发第一次AI寒冬的最重要事件之一，而另一事件则是联结主义的衰落。

Fall of Connectioni

联结主义的衰落

In 1969, Minsky and Papert’s book Perceptrons was published. It was a harsh critique on Rosenblatt’s perceptrons. Minsky and Papert proved that perceptrons could only be trained to solve linear separable problems. For instance, one of the most dooming examples of non-linear separable problems is the exclusive OR (XOR). For the network to successfully solve the XOR problem, its output can only be true if one of its inputs is true, but not both. This represented a big hit for the connectionists, who believe AI could be best achieved by mimicking the brain. Minsky and Papert knew that multiple layers would be able to solve this problem, but there was no algorithm to train such a network. It took 17 years until such an algorithm, now known as backpropagation, was devised. Only later on, was it discovered that the backpropagation algorithm had been discovered before. Indeed, it turned out that the backpropagation algorithm had been invented before Perceptrons was even published.

1969年，明斯基和帕佩尔出版了《Perceptrons》一书。该书对罗森布拉特的感知器进行了严厉的批评。明斯基和帕佩尔证明了，感知器只能被训练用于解决线性可分问题。例如，非线性可分问题中最令人感到遗憾的一个例子是异或（XOR）问题。一个能真正解决异或问题的网络，只有在一个输入为真，一个输入为假时，其输出才能为真。这对于联结主义者而言是一个巨大的打击，因为他们认为模仿大脑（的结构）是实现AI的最佳方式。明斯基和帕佩尔知道多层（且带有神经激活单元的）感知器能够解决这一问题，但是当时却没有算法可以用来训练这种网络。在17年之后，一种叫作“反向传播”的算法才被设计出来用于解决这个问题。但随后人们发现，其实在很早之前反向传播算法就已经存在了。更确切的讲，在《Perceptrons》一书出版之前，反向传播算法就已经被发明了出来。（只是没有人知道罢了）

The Lighthill report and its Consequences

莱特希尔报告及其后果

The Lighthill report (published in 1973) was an evaluation of the current state of AI at that time written for the British Science Research Council. The report came to the conclusion that the promises of AI researchers were exaggerated “in no part of the field have discoveries made so far produced the major impact that was then promised.” Though it pointed out that the most disappointing area of research had been machine translation, “… where enormous sums have been spent with very little useful result…” James Lighthill, the author of the report, thought that the failure to defeat the “combinatorial explosion” was at the heart of the issue. The term combinatorial explosion he refers to is a well-known problem in search spaces, like trees, where the number of nodes increases exponentially when going down the tree. For example, in a game like chess, Shannon demonstrated that the number of possible games increases from 20 with the first move to 400 with the second move, and by the 5th move there are already 4,865,609 possibilities, thus representing a combinatorial explosion.

莱特希尔报告（1973年出版）为英国科学研究理事会所撰写，它对人工智能的发展现状进行了评估。该报告得出结论说，人工智能研究人员夸大了他们的承诺，“迄今为止，在该领域（人工智能）的任何一个分支中，都没有出现他们当时承诺的重大成果”。同时，报告还指出最令人失望的研究领域是机器翻译，“……（它）花费了大量的资金，但成果极少”。报告的作者詹姆斯莱特希尔认为，未能打败“组合”是问题的核心所在。他提到的“组合”这个术语在空间搜索中是一个众所周知的问题。例如在树中，随着树由上到下逐渐变深，节点的数量呈指数级增长。在像国际象棋这样的游戏中，Shannon证明了游戏的可能状态数量从第一次时的20增加到第二次时的400，而在第五次时，游戏的可能状态已经变成了09种。这就是所谓的组合。

There was a lot of critici of the Lighthill report at the time and even a debate filmed for the BBC that unfortunately never got televised. Comparisons to other fields of science were made and comments made that one should not expect results that fast. Though the comments were arguably correct, the report had an effect. After the report, the UK government cut funding for all but two universities involved in research in this field, and it started a wave that swept throughout Europe and even had an impact on the U.S.

当时有很多人批评了莱特希尔的报告；BBC甚至为此拍摄了一场辩论，可惜最后没有播出。一些评论说，和其他科学领域进行比较后就能明白，人们不应该期待在短时间内就能取得成果。尽管这些评论可以说是正确的，但莱特希尔报告仍然得到了执行。报告发布后，英国削减了该领域除两所大学以外的所有资金，这一举措引发的浪潮席卷了整个欧洲，甚至对美国产生了影响。

The first AI Winter

第一次AI寒冬

Several circumstances combined to create the first AI winter. In the beginning, enthusia grew quickly about the potential of this new field with high amounts of optimistic press coverage. Then, disappointments in machine translation created a quiet era. Followed by Minsky and Pappert putting forward obstacles that impeded the progress of perceptrons. Finally, resulting in the instruction to create a realistic evaluation of the field, the Lighthill report came. With the arrival of the Lighthill report, the first winter started around 1973. The report had an effect on funding thus research on AI became difficult. DARPA (Defense Advanced Research Projects Agency) started funding more applied AI projects and less fundamental work. The AI winter lasted for a few years, but in the early 1980s, the field of AI experienced another high.

以上的几种状况相结合之后，第一个AI寒冬到来了。最初，人们对这一新领域的潜力热情似火，并制造了大量乐观的新闻报道。随后，机器翻译的失败导致了一个萧条的年代。然后，明斯基和帕普尔特对感知器的批评阻碍了它的发展。最后，莱特希尔报告导致相关机构开始对该领域进行更为现实的评估。随着莱特希尔报告的发布，第一次AI寒冬在1973年左右降临。该报告影响了相关项目的资金支持，因此对于AI的研究变得更加困难。DARPA（国防高级研究计划局，Defense Advanced Research Projects Agency）开始将更多的资金投向AI的应用研究，同时减少了对AI基础研究的资助。AI寒冬持续了好几年，但是在20世纪80年代初，人工智能领域又开启了另一个。

Events Leading to the Second AI Winter

导致第二次AI寒冬的事件

After the effects of the first AI winter had begun to decline, a new era of AI began to start. This time a lot more effort was focused on creating commercial products. Additionally, large conferences, like AAAI started in the early 1980s and experienced a rapid increase in tickets sold. The general industry and government officials alike started showing a renewed interest in AI technology.

在第一次AI寒冬的影响开始下降之后，又一个人工智能的新时代开始了。这一次，更多的注意力投向了商业产品的生产。此外，在20世纪80年代初期，像AAAI这样大型的会议门票销售数量急剧增长。一般行业和官员都开始对人工智能技术重新产生了兴趣。

At the heart of the commercialization of AI were expert systems. These systems were handcrafted by surveying experts and creating “if-then” rule sets accordingly. This method is called the “top-down” approach to AI with many believing that expert knowledge was the best way to create AI. These systems were implemented in fields like financial planning, medical diagnosis, geological exploration, and microelectronic circuit design.

这一次人工智能商业化的核心是专家系统。这些专家系统由相关专家通过“if-then”的规则集亲手编写而成。这种创建AI的方法被称为“自上而下的”，而且许多人认为专家知识是创建AI的最佳方式。在诸如财务规划、医疗诊断、地质勘探和微电子电路设计等领域，专家系统都得到了应用。

The magazine Business Week joined the hype and published the headline “AI It’s Here” in 1984. Similarly, many companies made extraordinary claims like “We’ve built a better brain” and declared that “[I]t is now possible to program human knowledge and experience into a computer … Artificial intelligence has finally come of age.”

《商业周刊》加入到大肆宣传的狂欢之中；1984年，它刊发了题为《AI：它在这》的头条文章。类似的，许多都夸大宣传说“我们已经制造了一个更好的大脑”，并且宣称“将人类的知识和经验编入到计算机之中已经成为可能……人工智能终于成熟了。”

Fears of an upcoming Winter

凌冬将至的恐惧

As the hype regarding AI increased, researchers feared that the field might not deliver the expected results. At a panel called “The Dark Ages of AI—Can we avoid or survive them?” at the 1984 AAAI conference scientists discussed if an upcoming AI winter could be prevented.

随着AI炒作的与日俱增，研究人员开始担心该领域可能无法实现预期的结果。1984年AAAI会议，在一个名为“人工智能的黑暗时代——我们可以避免它们或者幸存下来吗？”的小组讨论中，科学家们讨论了是否可以预防即将到来的AI寒冬。

“This unease is due to the worry that perhaps expectations about AI are too high, and that this will eventually result in disaster. I think it is important that we take steps to make sure the AI winter doesn’t happen […].”

“这种不安源于一种忧虑，人们对于人工智能的期望太高了，这终会导致灾难。我认为重要的是，我们应当采取行动以确保AI寒冬不会再次发生……”

The fear was that funding would once again dry up, when unrealistic expectations could not be fulfilled. This fear proved to be correct.

令人担心的是，当这些不切实际的期望无法被实现时，相关资金将会再次枯竭。这种担心被证明是正确的。

The second AI Winter

第二次AI寒冬

In the following years, the claims of what AI systems were capable of slowly had to face reality. The expert systems at the center of the revolution faced many issues. In 1984, John McCarthy criticized expert systems because they lacked common sense and knowledge about their own limitations.

在接下来的几年里，关于人工智能系统到底能做什么的宣传慢慢地开始面对现实。位于革命中心的专家系统遇到了很多问题。1984年，约翰麦卡锡对专家系统进行批评说，它们缺少对自身局限性的常识和知识。

John McCarthy

He described the expert system MYCIN built to assist physicians. He then laid out a situation where a patient has Cholerae Vibrio in his intestines. When asked, the systems prescribed two weeks of tetracycline. This would most likely kill off all the bacteria, but by then the patient would already be dead. Additionally, many tasks were too complicated for engineers to design rules around them manually. Systems for vision and speech contained too many edge cases.

他拿为帮助医生而建立的专家系统MYCIN为例进行了说明。他向系统询问，如果一个病人身上有霍乱弧菌的话应该怎么办？系统告诉他可以使用两周的四环素。毫无疑问，这确实可以杀死所有的细菌，但到那时病人也已经死掉了。此外，许多任务过于复杂，工程师根本无法设计相关规则。像视觉系统和系统就包含了太多个例的情况。

Schwarz, Director of DARPA ISTO (Defense Advanced Research Projects Agency/Information Science and Technology Office) from 1987 to 1989 concluded that AI research has always had

1987年至1989年，DARPA ISTO（国防高级研究计划局/信息科学与技术办公室）主任Schwarz认为人工智能研究

“… very limited success in particular areas, followed immediately by failure to reach the broader goal at which these initial successes seem at first to hint…”.

“……在特定领域取得的成功非常有限，并且在那些它们最初暗示能够成功的领域未能达成更为广泛的目标。”

This led to a decrease in funding in AI research. The general interest in AI declined as the expectations could not be met. At this time, many AI companies closed their doors. The AAAI conference that attracted over 6000 visitors in 1986 quickly decreased to just 2000 by 1991. Similarly, a decrease in AI-related articles starting in 1987 and reaching its lowest point in 1995 can be observed in The New York Times.

这最后导致了人工智能研究的资金开始变少。因为无法满足预期要求，人们普遍开始对人工智能失去兴趣。于是，许多人工智能在这时都破产了。1986年吸引了6000多人参会的AAAI会议在1991年就只有2000多名参会人员了。同样，从1987年开始，上AI相关的报道逐渐减少，并在1995年达到了最低点。