Will AI Replace Your Doctor?人工智能會(huì)取代醫(yī)生嗎？

馬克·西格爾/文　云天/譯

Fear of lawsuits might intimidate doctors and cause them to avoid going against artificial intelligence recommendations for treatment， even if their judgment tells them to do so.

出于對(duì)官司的恐懼，醫(yī)生可能被迫接受人工智能的治療建議，即便他們的判斷與人工智能的相悖。

Artificial intelligence can be a useful scientific tool， but it also could threaten a doctor’s essential role.

人工智能可以是一個(gè)實(shí)用的科學(xué)工具，但它也可能威脅到一名醫(yī)生的關(guān)鍵職責(zé)。

Medical school provided a similar kind of intelligence for me in that my brain was bombarded with a billion factoids1， which laid a tapestry2of information. Buried in my unconscious mind was the minutiae3 of nephrology， which I could bring to bear4 when I had a sick kidney patient， or the physiology of a failing heart when my patient’s lungs filled with fluid.

醫(yī)學(xué)院賦予了我類似的智能，我的大腦中存儲(chǔ)著十億條醫(yī)學(xué)事實(shí)陳述，它們編出一匹信息織錦。接診腎病患者時(shí)，我可以調(diào)用埋藏在我潛意識(shí)中的腎臟病學(xué)小知識(shí)；遇到雙肺充滿積液的患者，我的頭腦中又會(huì)出現(xiàn)有關(guān)心臟衰竭的生理學(xué)知識(shí)。

But the medical applications of computerized artificial intelligence are different. AI is based on a more precise pattern recognition5 from retinal analysis of diseases such as dia-betes or heart disease before it occurs， or even the kind of mutations of a glioma （one of the worst kinds of brain tumors） while surgery is still going on.

計(jì)算機(jī)化人工智能在醫(yī)療領(lǐng)域的應(yīng)用卻有所不同。人工智能基于一種更精確的模式識(shí)別技術(shù)。這種技術(shù)可以通過(guò)視網(wǎng)膜分析，在糖尿病、心臟病等疾病發(fā)生前識(shí)別出疾病，甚至還可以在手術(shù)過(guò)程中識(shí)別出神經(jīng)膠質(zhì)瘤（最嚴(yán)重的腦部腫瘤之一）的突變類型。

Recent studies also have examined pre-cancerous stem cells in the blood as well as other factors for AI to analyze that could help with diagnosis and treatment.

近來(lái)的研究還檢驗(yàn)了血液中的癌前干細(xì)胞，以及其他可供人工智能分析的、有助于疾病診療的因素。

Nonetheless， what AI will always lack is my clinical judgment formed from years of experience， not to mention my empathy for my patients. Increasingly， AI threatens that.

盡管如此，人工智能始終缺乏我從多年行醫(yī)經(jīng)驗(yàn)中形成的臨床判斷力，更不可能擁有我對(duì)病人的同情心。人工智能對(duì)同情心構(gòu)成的威脅甚至日益加劇。

Boundary between computer and doctor is easily blurred

計(jì)算機(jī)和醫(yī)生之間的界限很容易模糊

You have only to look at a study published in Nature Biomedical Engin-eering that looked at intraoperative video monitoring to instruct surgeons to understand that the boundary between doctor and computer is too easily blurred in a way that could intimidate or even threaten a surgeon’s abilities.

《自然·生物醫(yī)學(xué)工程》雜志上發(fā)表了一項(xiàng)關(guān)于術(shù)中視頻監(jiān)測(cè)技術(shù)指導(dǎo)外科醫(yī)生進(jìn)行手術(shù)的研究，你只需看看這項(xiàng)研究就會(huì)明白醫(yī)生和計(jì)算機(jī)之間的界限太容易模糊，以至于計(jì)算機(jī)可能會(huì)脅迫醫(yī)生，甚至有損醫(yī)生的能力。

And what about malpractice risk？ If you are a radiologist， a dermatologist or a surgeon who decides to disagree with your AI feed， based on years of clinical judgment and experience， and you end up being proven wrong in retrospect6， what is to prevent your patient from suing you and using the AI recommendations as evidence？

那么，是否存在治療失當(dāng)?shù)娘L(fēng)險(xiǎn)呢？如果你是一名放射科、皮膚科或外科醫(yī)生，根據(jù)自己多年的臨床判斷和經(jīng)驗(yàn)，決定拒絕采納人工智能的建議。事后證明你判斷有誤，那么如何防止你的病人以人工智能的建議為證據(jù)起訴你呢？

This might well intimidate doctors from going against AI recommendations for treatment， even if their judgment tells them to do so.

這很可能會(huì)讓醫(yī)生不敢違背人工智能的治療建議，即便他們的判斷與人工智能的相悖。

Consider that AI， when applied to clinical medicine， can give you only general answers. It cannot know the nuances7 of your case or history.

人工智能應(yīng)用于臨床醫(yī)學(xué)時(shí)，只能給醫(yī)生普遍適用的答案——它不可能了解醫(yī)生手中的病例或病史與其他的有何細(xì)微差別。

ChatGPT is a popular new AI bot that answers users’ questions. Patients already are using it for medical advice.

ChatGPT是一款很受歡迎的新型人工智能機(jī)器人，可以回答用戶的問題。病人已經(jīng)在向它尋求醫(yī)療建議了。

As Dr. Isaac Kohane， chair of the Department of Biomedical Informatics at Harvard， told The New England Journal of Medicine， “Now with these large language models like ChatGPT， you see patients going to and asking questions that they’ve wanted to ask their doctors—because， as is commonly the case， you forget things when you go to the doctor， perhaps because you’re stressed， and because， unfortunately， doctors don’t have that much time.”

正如哈佛大學(xué)生物醫(yī)學(xué)信息學(xué)系主任艾薩克·科漢博士在接受《新英格蘭醫(yī)學(xué)雜志》的采訪時(shí)所說(shuō)，“現(xiàn)在有了像ChatGPT這樣的大型語(yǔ)言模型，你會(huì)看到病人向這類語(yǔ)言模型詢問原本想問醫(yī)生的問題——因?yàn)椋ǔＧ闆r下，你去看醫(yī)生時(shí)容易忘記要說(shuō)的話，也許是因?yàn)槟憬箲]不安，也許是因?yàn)槟愫懿恍业嘏錾厢t(yī)生沒有那么多時(shí)間。”

I’m concerned about how patients will use advice from AI

我對(duì)患者將如何應(yīng)用人工智能的建議感到擔(dān)心

Kohane is excited about this advancement， but I am deeply concerned about it. While he is right that my availability in the office for face time with my patients is limited， especially because of electronic health records documentation requirements， the solution is definitely not after-visit consultations with artificial intelligence， which could easily provide information that misleads rather than helps a patient.

科漢為人工智能的進(jìn)步感到興奮，但我卻對(duì)此深感憂慮。他說(shuō)得沒錯(cuò)，我在診室與病人面對(duì)面交流的時(shí)間有限，特別是由于電子健康檔案存在諸多要求，但解決方案絕對(duì)不是讓病人在就診后咨詢?nèi)斯ぶ悄埽驗(yàn)槿斯ぶ悄芎苋菀滋峁┱`導(dǎo)而非幫助病人的信息。

In the same New England Journal of Medicine article， another AI expert， Dr. Maia Hightower， chief digital and technology officer at University of Chicago Medicine， pointed out the growing role of AI as an administrative tool in the opaque8 interface9 among doctors， patients and insurance companies.

在同一篇《新英格蘭醫(yī)學(xué)雜志》文章中，另一位人工智能專家、芝加哥大學(xué)醫(yī)學(xué)中心首席數(shù)字和技術(shù)官馬婭·海托華博士指出，在醫(yī)生、患者和保險(xiǎn)公司之間非透明的相互溝通中，人工智能作為一種管理工具的作用越來(lái)越大。

“So in order to communicate with payers， with our insurance companies， we’ll often have bots or automation that transfers information from the health system to the insurance company and back，” Hightower said. “In the case of insurance companies， we know that they often will use AI algorithms for prior authorization of procedures， whether or not to cover a particular medication or test. And in those cases， there isn’t much transparency on our side as a provider organization.”

“因此，為了與付款人和保險(xiǎn)公司溝通，我們通常會(huì)使用機(jī)器人程序或自動(dòng)化技術(shù)將信息從醫(yī)療系統(tǒng)傳輸?shù)奖ｋU(xiǎn)公司，然后再傳回。”海托華說(shuō)，“就保險(xiǎn)公司而言，我們知道他們通常會(huì)使用人工智能算法來(lái)處理預(yù)授權(quán)程序，無(wú)論最終是否承保特定的藥物或檢查。在這些情況下，作為醫(yī)療服務(wù)提供者，我們這邊的透明度并不高。”

As a practicing internist， I have a big problem with this and can envision a future where fights for insurance coverage become even more escalated than they are already—and where personalized medicine is replaced by algorithms. What’s to stop insurance companies from replacing me with a cheaper， more predictable AI robot， who practices some of the science but none of the art of medicine？

作為一名執(zhí)業(yè)內(nèi)科醫(yī)生，我對(duì)此有很大的意見。我可以預(yù)見，未來(lái)有關(guān)承保范圍的爭(zhēng)斗將比現(xiàn)在更加激烈，個(gè)性化醫(yī)療將被算法所取代。有什么能阻止保險(xiǎn)公司用一個(gè)更便宜、更可預(yù)測(cè)的人工智能機(jī)器人來(lái)取代我呢？這個(gè)機(jī)器人的確會(huì)進(jìn)行一些科學(xué)實(shí)踐，但它不懂醫(yī)學(xué)的藝術(shù)。

At best， AI will be like the automatic pilot in a commercial jet. It can help fly the plane， but passengers still want a human pilot in the cockpit ready to take control when needed.

人工智能充其量就如商用飛機(jī)上的自動(dòng)駕駛儀。這種儀器可以幫忙駕駛飛機(jī)，但乘客仍然希望駕駛艙內(nèi)有一名人類飛行員，隨時(shí)準(zhǔn)備著在需要時(shí)控制飛機(jī)。

（譯者為“《英語(yǔ)世界》杯”翻譯大賽獲獎(jiǎng)?wù)撸?/p>

1 factoid仿真陳述（很可能不真實(shí)但使人信以為真）；（尤指有關(guān)不太重要事情的）有趣消息。? 2 tapestry掛毯；織錦。? 3 minutia微小的細(xì)節(jié)（復(fù)數(shù)minutiae）。? 4 bring to bear應(yīng)用，使用。

5 pattern recognition模式識(shí)別，指計(jì)算機(jī)根據(jù)特征、屬性或結(jié)構(gòu)，確定一個(gè)物體、行為或現(xiàn)象屬于某個(gè)模式類別或模式個(gè)體的過(guò)程。一般分為數(shù)據(jù)獲取、基元分析、特征抽取、對(duì)象分類和后續(xù)處理等步驟。

6 in retrospect回顧。? 7 nuance細(xì)微差別。

8 opaque不透明的。? ?9 interface相互聯(lián)系；相互作用。

Thought-Provoking Quotes About AI

Nobody phrases it this way， but I think that artificial intelligence is almost a humanities discipline. It is really an attempt to understand human intelligence and human cognition.

—Sebastian Thrun

As society and the problems that face it become more and more complex and machines become more and more intelligent， people will let machines make more of their decisions for them， simply because machine-made decisions will bring better results than man-made ones. Eventually a stage may be reached at which the decisions necessary to keep the system running will be so complex that human beings will be incapable of making them intelligently. At that stage the machines will be in effective control.

—Ted Kaczynski

It is customary to offer a grain of comfort， in the form of a statement that some peculiarly human characteristic could never be imitated by a machine. I cannot offer any such comfort， for I believe that no such bounds can be set.

—Alan Turing