机器人三定律

机器人三定律（英语：Three Laws of Robotics）是科幻小说家艾萨克·阿西莫夫（Isaac Asimov）在他的机器人相关作品和其他机器人相关小说中为机器人设定的行为准则，是阿西莫夫除“心理史学”（Psychohistory）外另一个著名的虚构学说。

定律

第一法则

机器人不得伤害人类，或坐视人类受到伤害；

第二法则

机器人必须服从人类命令，除非命令与第一法则发生冲突；

第三法则

在不违背第一或第二法则之下，机器人可以保护自己。

在1985年《机器人与帝国》中，艾西莫夫将三大法则扩张为四大法则：^[1]

第零法则

机器人不得伤害整体人类，或坐视整体人类受到伤害；

第一法则

除非违背第零法则，否则机器人不得伤害人类，或坐视人类受到伤害；

第二法则

机器人必须服从人类命令，除非命令与第零或第一法则发生冲突；

第三法则

在不违背第零、第一或第二法则之下，机器人可以保护自己。

简介

“机器人三定律”在艾萨克·阿西莫夫于1942年发表的作品《转圈圈》（Runaround，《我，机械人》（I, Robot）中的一个短篇）中第一次明确提出，载于小说里虚构的《西元2058年第56版机器人手册》中^[2]，并且成为他的很多小说，包含基地系列小说中机器人的行为准则和故事发展的线索。机器人被设计为遵守这些准则，违反准则会导致机器人受到不可恢复的心理损坏。但是在某些场合，这样的损坏是不可避免的。在两个人互相造成伤害时，机器人不能任人受到伤害而无所作为，但是这会造成对另一个人的伤害，在一些小说中这造成了机器人的自毁。

他在小说中提出了三定律的很多变体。这些对机器人行为准则的修改成为他的很多故事的主线。在《消失无踪》（Little Lost Robot，《我，机械人》中的一个短篇）中为了避免机器人无谓地阻止人类进行低危险性工作，第一定律中的后半部分被去掉了，但是也产生了一些其他的问题^[2]。在《机械人之梦》（Robot Dreams，同名小说中的短篇）中，有个机器人因为正子脑出现不寻常碎形结构而进入无意识与梦境状态；梦中第一及第二法则不复存在，而第三法则被修改为“机器人必须保护自身的存在”^[3]。在《机器人与帝国》中，有些机器人因为系统中“人类的定义”被修改（说话没有该星球的口音就不是人类）而攻击人类^[4]。在1999年上映的科幻电影《双百人》（Bicentennial Man，原为短篇小说，后改编为中篇小说《正子人》）中，一个机器人为了成为人，将自己的机械部分逐步替换成组织，并且在最后用自己的贡献和牺牲换来了人类的认同和保护机器人的法律的产生。^[5]

三定律在科幻小说中大放光彩，在一些其他作者的科幻小说中的机器人也遵守这三条定律。同时，三定律也具有一定的现实意义，在三定律基础上建立新兴学科“机械伦理学”旨在研究人类和机械之间的关系。但是现存的机械人并未有足够的智能去分辨人类和危险等概念，因此三定律并未实际应用在机械人上面；或许要等人工智慧有显著进展才有可能，只不过就算AI智能可达到人类水平，内在伦理的复杂性加上定律有赖于文化与情境，将成为制定“机器人约束”的莫大挑战。^[6]

修改和扩充

在此之后，不少其他作者添加或修改定律，例如有更改第三定律，让第二定律不再置于第三定律之前（让人类不能命令机械人自灭），另一些则是让机械人要一直认为自己是机械人（某些故事中的机械人把自己定义为人类，从而无视三定律）^[7]。因此出现了很多不同的第四、第五甚至第六定律。^[8]^[9]

2020年，Frank Pasquale提出“机器人新律”：1.机器人系统与人工智慧应为补充与辅助专业人士的角色，而非取代专业人士。2. 机器人系统与AI不应该假冒人。3. 机器人系统和人工智能不应该加剧零和博弈的军备竞赛。4.机器人系统与AI必须随时标示其创造者、控制者及拥有者的身分^[10]。Frank Pasquale希望透过“决策民主化”的方法来探索人类与AI如何合作以保障未来繁荣，而不是将希望寄托在企业身上。 ^[11]

评价

中国武汉大学的哲学家Chris Stokes认为所谓“机器人三定律”由于语言上的歧义和道德方面的问题，在现实世界中不起作用，在编程中也难以实现^[12]。

参考

参考来源

^ Isaac Asimov. BBC. [11 November 2010]. （原始内容存档于2010-01-10）.
^ ^2.0 ^2.1 Asimov, Isaac. Runaround. I, Robot The Isaac Asimov Collection. New York City: Doubleday. 1950: 40. ISBN 978-0-385-42304-5 （英语）. This is an exact transcription of the laws. They also appear in the front of the book, and in both places there is no "to" in the 2nd law.
^ Asimov, Isaac. Robot Dreams. 1986 [11 November 2010]. “But you quote it in incomplete fashion. The Third Law is ‘A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.’ ” “Yes, Dr. Calvin. That is the Third Law in reality, but in my dream, the Law ended with the word ‘existence’. There was no mention of the First or Second Law.”
^ Asimov, Isaac. Robots and Empire. Doubleday books. 1985: 151. ISBN 978-0-385-19092-3. although the woman looked as human as Daneel did, she was just as nonhuman
^ Asimov, Isaac. The Complete Stories, Volume 2. Doubleday. 1992: 568–604. ISBN 978-0-385-42078-5.
^ Murphy, Robin; Woods, David D. Beyond Asimov: The Three Laws of Responsible Robotics (PDF). IEEE Intelligent Systems. July 2009, 24 (4): 14–20 [2023-12-08]. S2CID 3165389. doi:10.1109/mis.2009.69. （原始内容存档 (PDF)于2023-04-09）.
^ Don D'Ammassa. Allen, Roger MacBride. Encyclopedia of science fiction. Infobase Publishing. 2005: 7. ISBN 978-0-8160-5924-9.
^ Lawful Little Country: The Bulgarian Laws of Robotics | The Portal. [2023-12-08]. （原始内容存档于2011-10-06）.
^ Ashrafian, Hutan. AIonAI: A Humanitarian Law of Artificial Intelligence and Robotics. Science and Engineering Ethics. 2014, 21 (1): 29–40. PMID 24414678. S2CID 2821971. doi:10.1007/s11948-013-9513-9.
^ Frank Pasquale. 二十一世紀機器人新律：如何打造有AI參與的理想社會？. 由李, 姿仪翻译. 台湾: 左岸文化. 2023-03-08. ISBN 9786267209271.
^ Frank Pasquale. New Laws of Robotics: Defending Human Expertise in the Age of AI. Belknap Press: An Imprint of Harvard University Press. 2020-10-27. ISBN 978-0674975224.
^ The Three Laws of Robotics Have Failed the Robots. Mind Matters. 2019-09-28 [2022-12-07]. （原始内容存档于2022-12-07）. Chris Stokes, a philosopher at Wuhan University in China, says, “Many computer engineers use the three laws as a tool for how they think about programming.” But the trouble is, they don’t work.
He explains in an open-access paper:The First Law fails because of ambiguity in language, and because of complicated ethical problems that are too complex to have a simple yes or no answer.The Second Law fails because of the unethical nature of having a law that requires sentient beings to remain as slaves.The Third Law fails because it results in a permanent social stratification, with the vast amount of potential exploitation built into this system of laws.

[BBCAsimov-1] Isaac Asimov. BBC. [11 November 2010]. （原始内容存档于2010-01-10）.

[IROBOT-2] 2.0 ^2.1 Asimov, Isaac. Runaround. I, Robot The Isaac Asimov Collection. New York City: Doubleday. 1950: 40. ISBN 978-0-385-42304-5 （英语）. This is an exact transcription of the laws. They also appear in the front of the book, and in both places there is no "to" in the 2nd law.

[RDA1-3] Asimov, Isaac. Robot Dreams. 1986 [11 November 2010]. “But you quote it in incomplete fashion. The Third Law is ‘A robot must protect its own existence as long as such protection does not conflict with the First or Second Law.’ ” “Yes, Dr. Calvin. That is the Third Law in reality, but in my dream, the Law ended with the word ‘existence’. There was no mention of the First or Second Law.”

[ROBANDEMP1-4] Asimov, Isaac. Robots and Empire. Doubleday books. 1985: 151. ISBN 978-0-385-19092-3. although the woman looked as human as Daneel did, she was just as nonhuman

[5] Asimov, Isaac. The Complete Stories, Volume 2. Doubleday. 1992: 568–604. ISBN 978-0-385-42078-5.

[6] Murphy, Robin; Woods, David D. Beyond Asimov: The Three Laws of Responsible Robotics (PDF). IEEE Intelligent Systems. July 2009, 24 (4): 14–20 [2023-12-08]. S2CID 3165389. doi:10.1109/mis.2009.69. （原始内容存档 (PDF)于2023-04-09）.

[EoSF1-7] Don D'Ammassa. Allen, Roger MacBride. Encyclopedia of science fiction. Infobase Publishing. 2005: 7. ISBN 978-0-8160-5924-9.

[8] Lawful Little Country: The Bulgarian Laws of Robotics | The Portal. [2023-12-08]. （原始内容存档于2011-10-06）.

[9] Ashrafian, Hutan. AIonAI: A Humanitarian Law of Artificial Intelligence and Robotics. Science and Engineering Ethics. 2014, 21 (1): 29–40. PMID 24414678. S2CID 2821971. doi:10.1007/s11948-013-9513-9.

[10] Frank Pasquale. 二十一世紀機器人新律：如何打造有AI參與的理想社會？. 由李, 姿仪翻译. 台湾: 左岸文化. 2023-03-08. ISBN 9786267209271.

[11] Frank Pasquale. New Laws of Robotics: Defending Human Expertise in the Age of AI. Belknap Press: An Imprint of Harvard University Press. 2020-10-27. ISBN 978-0674975224.

[mindmatters-12] The Three Laws of Robotics Have Failed the Robots. Mind Matters. 2019-09-28 [2022-12-07]. （原始内容存档于2022-12-07）. Chris Stokes, a philosopher at Wuhan University in China, says, “Many computer engineers use the three laws as a tool for how they think about programming.” But the trouble is, they don’t work.
He explains in an open-access paper:The First Law fails because of ambiguity in language, and because of complicated ethical problems that are too complex to have a simple yes or no answer.The Second Law fails because of the unethical nature of having a law that requires sentient beings to remain as slaves.The Third Law fails because it results in a permanent social stratification, with the vast amount of potential exploitation built into this system of laws.

[1]

[2]

[3]

[4]

[5]

[6]

[7]

[8]

[9]

[10]

[11]

[12]