经济学人 | 如果有朝一日，人工智能获得诺贝尔医学奖.......
1
2
听力|精读|翻译|词组
Rage against the machine
横眉冷对计算机
英文部分选自经济学人20210702期What If版块
If an AI won the Nobel prize for medicine
如果有朝一日，人工智能获得诺贝尔医学奖......
Rage against the machine
横眉冷对计算机
Should the greatest prize in medical research really be awarded to a non-human? An imagined scenario from 2036
医学研究的最高奖真的应该颁发给非人类么？一段对2036年的设想
Editor’s note: This year What If?, our annual collection of scenarios, considers the future of health. Each of these stories is fiction, but grounded in historical fact, current speculation and real science. They do not present a unified narrative but are set in different possible futures
编者按:年度栏目“如果……会怎样?”旨在探讨一些假设场景，今年我们聚焦健康领域的未来。这里面的每一个故事虽为虚构，但都建立在历史事实、当前猜测以及真实科学的基础之上。这些故事各自成篇，设定于不同的未来世界。
IT WAS A scene that the Nobel committee had dearly hoped to avoid. As the recipients of this year’s prizes filed into the Stockholm Concert Hall to take their seats, dozens of protesters, including several former laureates, clashed with police in the streets outside. They had gathered to express their opposition to the unprecedented decision to award the Nobel prize in physiology or medicine to an artificial intelligence.
点击显示翻译
▼
今年的诺奖得主们在斯德哥尔摩音乐厅就坐时，外面街道上大批的抗议者（包括几位前诺奖得主）与警方发生了冲突。他们聚集在一起，反对将诺贝尔生理学奖或医学奖颁给人工智能这一史无前例的决定。这是诺贝尔奖委员会过去一直极力避免的一幕。
注释：
file in(to something or some place) To move or process into a place or thing in a line.
The committee’s citation recognised YULYA—the nickname of a machine-learning system officially known as System for Automated Lymphoma Diagnosis—as the discoverer of ancillary vulnerability, a mechanism whereby specific pairs of antibiotics, working in tandem, can prove effective against bacteria that are otherwise resistant. The committee estimates that in the 18 months since the discovery, which occurred when the death rate associated with the failure of existing antibiotics had risen to around 2.5m a year, YULYA’s work has saved around 4m lives, both through direct treatment of infections and by allowing the resumption of surgical procedures, including caesarean sections, that were considered too dangerous without antibiotics.
点击显示翻译
▼
诺奖委员会的颁奖词中，肯定了YULYA（一个机器学习系统的昵称，官方名称为“自动化淋巴瘤诊断系统”）在发现附属脆弱性（ancillary vulnerability）方面作出的贡献。通过附属脆弱性机制，一些特定的抗生素组合在协同作用下可以有效解决细菌耐药性问题。委员会估算，由于当时全球每年与抗生素失效相关的死亡人数已达约250万，自YULYA发现附属脆弱性的18个月内，通过直接治疗感染，以及帮助恢复包括剖宫产在内的一系列（若缺乏有效抗生素则极其危险的）手术，它已经拯救了大约400万条生命。
注释：
1. citation：A formal statement of the accomplishments of one being honored with an academic degree.
2. Lymphoma: 淋巴瘤
3. caesarean sections剖宫产
Bringing to an end the greatest global public-health crisis since the coronavirus pandemic of 2020-22 would, you might have thought, be considered qualification enough for anyone, whether human or machine, to win the Nobel prize. But the decision has proved hugely controversial. Though the statutes of the Nobel Foundation have historically been interpreted as implying that only a human can win the award, another of its dictates was deemed to take precedence: recognition for having “conferred the greatest benefit to humankind” in the preceding year. Another factor behind the break with tradition was a demographic shift in the prize committee. When two of the committee’s five members succumbed to bacterial infections last year, younger replacements were elected, both of whom happened to have used machine-learning systems in their doctoral research.
点击显示翻译
▼
你也许认为，任何个体（无论是人还是机器），如果能够终结自2020-2022年新冠疫情以来最严重的全球公共卫生危机，都有足够的资格拿下诺贝尔奖。然而事实证明，这个决定极具争议。尽管诺贝尔奖基金会的章程一直被解读为只有人类可以获奖，但是其中另一项规定似乎更需优先考虑：表彰上一年度“为人类作出杰出贡献”。打破传统的另一个原因是诺奖委员会里人口结构的变化。前一年中，五名时任委员中有两位因细菌感染离世，而两名更年轻的新当选评委在他们的博士研究中均使用过机器学习。
YULYA was originally built to tackle a different problem: finding more effective cancer treatments. One of the world’s most advanced causal nets, it is one of a new generation of artificial-intelligence systems combining the pattern-recognition skills of conventional “deep” neural networks with the ability to distinguish causation from mere correlation. By examining records from patient databases, in conjunction with a corpus of papers from medical journals and historical data from pharmaceutical companies, it sought to identify the patterns of symptoms that led to the most severe outcomes, in order to diagnose them earlier. It was also programmed to evaluate the effectiveness of different treatments, including combinations of treatments, in order to suggest new therapeutic regimens that could be tested in patients.
点击显示翻译
▼
创造YULYA的初衷是为了解决另一个问题：寻找更有效的癌症疗法。作为世界上最先进的因果性网络之一，新一代人工智能系统YULYA结合了传统 “深度 ”神经网络的模式识别技能和区分因果关系与单纯关联性的能力。通过检查病患数据库中的记录，结合医学期刊的论文库和制药公司的历史数据，该系统试图识别出导致最严重后果的症状模式，从而尽早诊断。YULYA还被用来评估不同治疗方法的有效性，包括组合疗法，以便找到可以在病人身上测试的新治疗方案。
Its focus shifted, however, when a software upgrade in 2034 accidentally gave it access to all recent papers in medical journals, rather than just those associated with cancer. YULYA duly began to crunch data relating to antimicrobial resistance, which accounted for a steadily growing proportion of medical-research papers as the crisis intensified. At first, its requests for more data in specific areas, and suggestions for new approaches to treatment, were thought to be errors, because they did not relate to cancer. Then YULYA’s operators realised what had happened, and saw that it had used its reasoning capabilities to build a testable hypothesis: the forerunner of what would become ancillary vulnerability. It highlighted the data that would be needed to validate the hypothesis, including specific guidelines as to how it should be collected. “It amounted to a full-scale programme of research,” says Anisha Rai, one of YULYA’s creators.
点击显示翻译
▼
然而，2034年的一次软件升级意外地使YULYA能够访问所有最新医学期刊论文（不仅仅是那些与癌症有关的论文），这时该系统的研究重点也发生了转移。YULYA正式开始处理与抗菌素耐药性有关的数据，随着公共卫生危机加剧，抗菌素耐药性在医学研究论文中所占的比例也稳步上升。起初，YULYA提出更多特定领域的数据请求，并给出新治疗方法的建议，但都因与癌症无关而被人为是程序产生了错误。后来YULYA的操作人员意识到了这一状况，并发现YULYA利用自身的推理能力建立了一个可验证的假设：这就是附属脆弱性的前身。YULYA重点关注验证该假设所需的数据，包括如何收集数据的具体准则。YULYA的创造者之一Anisha Rai表示：“这相当于一个全面的研究计划。”
Under less exceptional circumstances, such trials might never have been authorised. Many funding bodies require scientists to lay bare the reasoning process of AI systems, in order to be sure that their recommendations do not lead to deadly conclusions. Dr Rai and her colleagues got funding for YULYA’s trial by playing down its role in suggesting the hypothesis. Only when the results showed promise did they publish YULYA’s original proposals.
点击显示翻译
▼
普遍情况下，这种试验可能永远不会获批。许多资助机构要求科学家披露人工智能系统的推理过程，确保它们的建议不会导致致命结果。Rai博士和她的同事们通过淡化YULYA在提出假说方面的作用，为YULYA的试验募得资金。一直到得出的结论有望成功的时候，他们才公布YULYA的最初建议。
That, in turn, led to a heated debate about whether YULYA, or its creators, deserved credit for the breakthrough. Dr Rai continues to insist that YULYA deserves sole credit, a position that has prompted the departure of several members of her original team in the past year. She even refused to go to Stockholm to receive the award on YULYA’s behalf from the queen of Sweden. “It’s not my prize,” she says.
点击显示翻译
▼
这反而引发了一场关于取得突破的功劳属于YULYA还是其创造者为的激烈争论。Rai博士仍坚持认为功劳只属于YULYA，这一立场导致她原团队中的几位成员在过去一年中相继离职。她甚至拒绝代表YULYA前往斯德哥尔摩接受瑞典女王的颁奖。她表示：“这份荣誉不属于我。”
AIs are commonly used to predict the onset of diseases like Alzheimer’s, make personalised treatment recommendations and enhance the diagnostic abilities of physicians. And the use of AI in drug discovery, in particular to help pharmaceutical companies wade through databases, is not new. In 2020 an algorithm developed at the Massachusetts Institute of Technology made headlines when it identified a new antibiotic. Dubbed halicin, after the computer in the film “2001: A Space Odyssey”, it proved to be effective against some resistant bacteria, but was limited in its scope. “Ancillary vulnerability makes halicin look like a homeopathic treatment, like a placebo,” says Una Científica, a researcher at the Houssay Institute in Buenos Aires.
点击显示翻译
▼
人工智能通常用于预测阿尔茨海默氏症等疾病的发作，提出个性化的治疗建议并提高医生的诊断能力。将人工智能用于药品研发——尤其是帮助制药公司处理数据——并不是什么新鲜事。2020年，麻省理工大学开发的一个算法因发现了一种新型抗生素而登上新闻头条。为致敬电影《2001太空漫游》中的超级计算机HAL，该抗生素被命名为halicin。它可以对抗一些耐药细菌，但范围有限。布宜诺斯艾利斯（阿根廷首都）奥赛（Houssay）研究所的研究员Una Científica表示：“附属脆弱性让halicin看起来像是一种顺势疗法，和安慰剂类似。”
注：
顺势疗法：是替代医学的一种。顺势疗法的理论基础是“同样的制剂治疗同类疾病”，意思是为了治疗某种疾病，需要使用一种能够在健康人中产生相同症状的药剂
Even so, the Nobel committee’s reference to YULYA’s “discovery” has angered those who see it as little more than a clever tool. “YULYA is an AI capable of winning a Nobel. That is not the same thing as an AI that’s capable of discovery,” says Hars Kritik of the European Robotics Institute in Prague. He argues that even the best AIs are only useful in specialised areas like drug design, where large quantities of data are married to well-defined metrics of success. Saying that they can make discoveries, he says, waving a placard outside the concert hall, is “flawed anthropomorphism”.
点击显示翻译
▼
即便如此，诺贝尔委员会将附属脆弱性称为YULYA的“发现”,让很多人大为恼火，他们认为YULYA只不过是一个聪明的工具。布拉格欧洲机器人研究所的Hars Kritik指出，“YULYA作为一种人工智能工具虽然可以获得诺奖，但并不代表她能够发现新事物。” 他认为，即使是最好的人工智能也只能在药物设计等专业领域发挥用处，在这些领域中，成功的明确衡量标准离不开大量的数据。他在音乐厅外一边挥舞着标语牌，一边说道：把人工智能说成能够发现新事物是一种“有缺陷的拟人论”。
Rightly or wrongly, YULYA is unlikely to be the last artificial intelligence to win a Nobel prize. Sources within the Nobel Foundation say that similar nominations have been received for prizes in physics and chemistry, as AI systems are used to search for new materials and chemical compounds suitable for use in batteries, solar panels and carbon-capture membranes. Given the chaos that erupted in Stockholm this week, however, the chances of an AI winning the Nobel peace prize seem rather more remote.
点击显示翻译
▼
无论对错，YULYA都不太可能是最后一个获得诺贝尔奖的人工智能工具。诺贝尔基金会内部人士称，物理学和化学奖也收到了类似的提名，因为人工智能系统被用于寻找适用于电池、太阳能电池板和碳捕获膜的新材料和化合物。然而，考虑到本周斯德哥尔摩爆发的混乱，人工智能获得诺贝尔和平奖的可能性似乎微乎其微。
翻译组：
Cathy，女，春都MTIer,专业翻译爱好者
Frank，男，帝都MTIer,爱看球，爱踢球，爱读经济学人
Diamond,男，搬砖工，理工科永久脑损伤，学习翻译的小学生
校对组：
奥莉，女，撸猫狂魔，强迫症患者
Rachel，学理工科，爱跳芭蕾，热爱文艺的非典型翻译
Mia，女，期待快点毕业的MTIer,有志从事翻译工作的小文青
感谢其他平台编辑作者：
B站地址：
https://space.bilibili.com/522030734
编辑：Daisy，勤勤恳恳工具人，B站客服的好朋友
知乎地址：
zhihu.com/people/damon0922
编辑：jill，会展经济，想跨界英专的英语学渣
微博地址：
https://weibo.com/u/2375202237
编辑：Erin，在校大学生一枚，TE爱好者
3
观点|评论|思考
文章开头的一幕很具有典型性，无论任何决定，在除了某大国外的其他地区几乎总是有抗议者，理论上而言这也很正常，绝大部分决定总是存在争议，人们也应该有表达相左意见的权利和途径。但这并不是我们今天讨论的重点。我的关注点有两个，所谓人工智能并非只是一台设备，还包括了设备的设计制造以及后期维护方；此外，对于此案例而言，还有一个更加重要的群体，那就是验证人工智能的建议是否正确的团队。前者文中已经提到了，似乎在文章作者看来，设计制造完一台人工智能设备后就万事大吉了，但显然这是外行人的观点，在现有的技术条件下（文章也说了这是根据历史事实和真实科学的推测，也就是说这不是毫无基础的科学幻想），所有所谓的人工智能或者类似产品，都需要进行维护，并不存在如电影里那般一劳永逸的可能性，而且维护团队的贡献绝对不亚于人工智能设备本身，所以人工智能是一个整体的概念，包括设计制造、设备本身和后期维护管理团队，单独凸显这台设备既是一种认知错误，更是一种荒唐可笑的举动。所以从这个角度来看，把奖颁给这台设备（或者这段程序）是无知的行为。
其二，目前科学家提出的假设的确大部分是由另外的科学家进行分析验证，来证明或者否定这个假设。但这些假设有一个大的前提，就是科学家拥有足够的背景知识和可靠性，不会提出一些超出逻辑认知范围的假设，因为我们是人类，我们的头脑中会有一个范围，一个属于常识逻辑的范围，就好像我们在楼顶不会跳下去，我们在马路上不会走向高速行驶的汽车，但是人工智能并不存在这种常识逻辑，它也许拥有预设的一些边缘，但是这种边缘总是有限的，不可能触及到全部的领域，所以人工智能在进行假设时也不可能拥有常识性逻辑边缘，这样一来，也许它的假设和想象力比人类会更加广阔，倘若是走围棋，那倒无妨，但是如果是文中提及的药品之类，这种便是毫无底线的假设，从这个意义上来说，对于假设的验证就像是人工智能的监护人，人工智能更像是一个毫无常识的永不长大的小婴儿，所以从这个意义上来说，如果人工智能的猜想对了，这个奖应该颁给它的监护人，而不是婴儿自己。
我想文章主要还是想表达人工智能对于科学贡献可能的突破性和重要性，单就这点来说是没有任何问题的，但是作者试图将这个概念融入一个假象的场景之中就略显愚笨了，这就好比网民们老是说的梅罗和贝马球王同场竞技，亦或是做一些违反常识和科学的假设，以试图达到吸引眼球、最终表达自己观点的目的，只能说水品不高、技术不行，还不如将前提去掉，说自己是科幻场景，如此大家讨论起来可能还会更加尽兴一些。
4
愿景
打造
独立思考 | 国际视野 | 英文学习
小组
现有经济学人打卡（讨论）群一个,每周必须完成一篇经济学人打卡，每个人只有一次机会进群，一旦没有完成打卡任务，请出后不能再次加入。如果您也有兴趣，可联系小编WeChat : Education0603。由于每天加小编人很多，为提高效率，大家添加小编，暗号“TE打卡群', 通过后，请做好以下三点，否则不回复，谢谢理解。
1.转发公号推文到英语学习群或者朋友圈
2.回答以下三个问题
为什么加入？你的初衷是什么？
你对经济学人的了解？
你希望离开的时候能得到什么？
3.加小编后做个简单的自我介绍，阅读群规能做到方可进群