創(chuàng)新工場(chǎng)董事長(zhǎng)兼首席執(zhí)行官

創(chuàng)新工場(chǎng)人工智能工程院院長(zhǎng)

新冠疫情加速中國(guó)自動(dòng)化大躍遷

2003年非典疫情期間，北京的實(shí)體商店全數(shù)關(guān)門歇業(yè)，這迫使一家零售商嘗試互聯(lián)網(wǎng)轉(zhuǎn)型，開拓新市場(chǎng)，這家零售商后來發(fā)展為當(dāng)今中國(guó)電子商務(wù)巨頭之一：京東。再把時(shí)間快進(jìn)到我們所在的2020年新冠疫情期間，京東大舉推行新型自動(dòng)化倉(cāng)儲(chǔ)，單日訂單處理能力超過150萬單，推動(dòng)銷售額飆升20%，還在武漢用上了機(jī)器人和無人機(jī)配送，還計(jì)劃于今年內(nèi)在各地開設(shè)1000家自動(dòng)化餐廳。

我們的世界正迎來疫情沖擊下的重大革新：自動(dòng)化、智能化和數(shù)字化。中國(guó)經(jīng)濟(jì)正在經(jīng)歷一場(chǎng)由自動(dòng)化驅(qū)動(dòng)的升級(jí)再造，大規(guī)模引入機(jī)器人技術(shù)來降低商業(yè)運(yùn)作中人與人的接觸；在線業(yè)務(wù)、算法和自動(dòng)化實(shí)現(xiàn)了降本提效及安全保障。雖然這樣的轉(zhuǎn)變?cè)诖舜我咔橹熬鸵寻l(fā)端，但疫情確實(shí)起到了加速的作用。同時(shí)我預(yù)測(cè)，自動(dòng)化技術(shù)短期內(nèi)可以用于保證社交安全距離，長(zhǎng)期而言將對(duì)商業(yè)運(yùn)行機(jī)制產(chǎn)生更廣泛、更深層次的影響。這一波后疫情時(shí)代的自動(dòng)化浪潮在中國(guó)如火如荼，或許也將于世界各地的商業(yè)環(huán)境開花結(jié)果。

過去人們以為要完成生活中的大事小情，就得與他人面對(duì)面互動(dòng)。經(jīng)歷疫情隔離后，他們發(fā)現(xiàn)其實(shí)并非如此。從醫(yī)療看護(hù)、餐飲、配送、制造、物流、運(yùn)輸和教育這類藍(lán)領(lǐng)屬性的職業(yè)，到金融、客服、銷售、人力資源、法律和會(huì)計(jì)這些白領(lǐng)屬性的工作，很大程度上都可以由軟件和機(jī)器人技術(shù)來完成。

從歷史角度看，當(dāng)經(jīng)濟(jì)下行周期恰逢新一波成熟技術(shù)的應(yīng)用落地，往往會(huì)掀起自動(dòng)化的熱潮。企業(yè)受到成本壓力后，會(huì)覺得有必要削減工作崗位和嘗試新技術(shù)。一旦企業(yè)成功引進(jìn)機(jī)器人來取代人類員工，老板們就不太可能走回頭路，因?yàn)闄C(jī)器人不會(huì)生病、不會(huì)曠工、不會(huì)因?yàn)橐獔?zhí)行危險(xiǎn)任務(wù)而要求漲薪。事實(shí)上，機(jī)器人可以說是疾病大流行期間取代高危崗位的最佳解決方案。難怪美國(guó)麻省理工學(xué)院的經(jīng)濟(jì)學(xué)家大衛(wèi)·奧特爾 (David Autor) 將新冠疫情稱為“自動(dòng)化推手”。

中國(guó)在發(fā)展自動(dòng)化經(jīng)濟(jì)方面有著得天獨(dú)厚的優(yōu)勢(shì)，雖然擁有龐大的勞動(dòng)力，但在過去20年間，中國(guó)的勞動(dòng)力成本增加了十倍，是越南的兩倍多。作為世界工廠，中國(guó)有十足的動(dòng)力實(shí)現(xiàn)智能制造，進(jìn)而從制造大國(guó)晉升為制造強(qiáng)國(guó)?，F(xiàn)在中國(guó)已經(jīng)是全球最大、增長(zhǎng)最快的工業(yè)機(jī)器人市場(chǎng)，2018年國(guó)內(nèi)的工業(yè)機(jī)器人市場(chǎng)規(guī)模激增21%，達(dá)到54億美元，占全球銷售額的三分之一，各行各業(yè)也在積極探索將機(jī)器人技術(shù)融入制造業(yè)的方方面面。

自動(dòng)化加速的趨勢(shì)還輻射到制造業(yè)以外的諸多領(lǐng)域。2月疫情在武漢迅速蔓延，規(guī)模龐大的火神山醫(yī)院在十天內(nèi)建成，成列的機(jī)器人在醫(yī)院內(nèi)奔走，進(jìn)行消毒和物資運(yùn)送，機(jī)器人的“身影”還出現(xiàn)在各地的學(xué)校、醫(yī)院和商業(yè)建筑中。上海機(jī)器人公司擎朗開發(fā)的一款無人車結(jié)合了激光雷達(dá)、計(jì)算機(jī)視覺、傳感器等技術(shù)，可執(zhí)行無接觸消毒任務(wù)。

前陣子我在北京的家中隔離時(shí)，網(wǎng)購(gòu)的包裹和食品都由小區(qū)的機(jī)器人運(yùn)送，它裝著輪子，外形敦實(shí)，很像《星球大戰(zhàn)》中的R2D2，送貨時(shí)先以無線呼叫電梯，自主導(dǎo)航至我家門口，然后撥打我的電話通知開門取件。我取走快遞后，機(jī)器人就會(huì)回到前臺(tái)等候下一個(gè)任務(wù)。

美團(tuán)外賣也推出了“無接觸”配送服務(wù)，將餐品送至指定的取餐點(diǎn)，外賣小哥和顧客無需互動(dòng)便可完成交易，美團(tuán)還在積極測(cè)試無人駕駛送餐車。微信也開發(fā)了一個(gè)系統(tǒng)，人們可以用手機(jī)瀏覽餐廳菜單、點(diǎn)餐和結(jié)賬，由人類或機(jī)器人將食物送上餐桌。穿梭于店內(nèi)的機(jī)器人服務(wù)員可能是為了隔離安全，也可能是為了炒噱頭，但未來它們或許會(huì)成為大多數(shù)普通餐廳的基礎(chǔ)配置（上流餐廳除外）。

不過短期來看，自動(dòng)化對(duì)白領(lǐng)類型的工作影響更大。雖然流水線這樣基礎(chǔ)的重復(fù)性工作容易實(shí)現(xiàn)自動(dòng)化，但多數(shù)手工技藝對(duì)機(jī)器人來說仍舊難以掌握。21世紀(jì)的“智能自動(dòng)化”將有別于20世紀(jì)的初級(jí)物理自動(dòng)化，因?yàn)槿缃竦臋C(jī)器人需要將機(jī)械工程和用于感知和精細(xì)動(dòng)作操控的人工智能技術(shù)模塊融為一體。而在疫情期間，白領(lǐng)員工居家上班，他們?cè)诰€上的一切活動(dòng)、承擔(dān)的所有任務(wù)都被轉(zhuǎn)化成數(shù)據(jù)，標(biāo)志著向機(jī)器全面接手邁出了一小步。這就是自疫情危機(jī)爆發(fā)以來，提供“機(jī)器人流程自動(dòng)化”的公司迎來一波銷售熱潮的原因。

作為身在中國(guó)的技術(shù)投資人，我第一手見證了這些技術(shù)在疫情期間落地應(yīng)用，舉例創(chuàng)新工場(chǎng)投資的追一科技公司開發(fā)了一款客服中心自動(dòng)化軟件。疫情期間，國(guó)內(nèi)一家大型銀行的信用卡部門部署使用了該軟件，每天和客戶的通話量高達(dá)35萬，相當(dāng)于1200名人工客服的工作量，這類對(duì)話機(jī)器人不僅降低了成本，還提高了客戶滿意度，增加了收入?，F(xiàn)在追一科技已將AI數(shù)字員工的落地應(yīng)用擴(kuò)大到智能營(yíng)銷、智能分析、智能培訓(xùn)、智能助理等諸多場(chǎng)景。

后疫情時(shí)代的商業(yè)環(huán)境新脈絡(luò)正在成形，一切可通過自動(dòng)化降本提效的運(yùn)營(yíng)環(huán)節(jié)都將實(shí)現(xiàn)自動(dòng)化，而自動(dòng)化過程壓縮人類崗位，不僅僅是為了利潤(rùn)和績(jī)效，更是為了健康和安全。機(jī)器人和軟件在疫情發(fā)生前便已展現(xiàn)出吸引力，如今這些技術(shù)更是有理由成為企業(yè)剛需——無論新冠肺炎疫苗是否會(huì)研發(fā)出來。

雖然一些工作因自動(dòng)化改變或消亡，但許多新的工作崗位也被創(chuàng)造出來，例如數(shù)據(jù)中心、5G設(shè)備、智能軟件等新的數(shù)字化基礎(chǔ)設(shè)施，軟件將需要人類員工加以研發(fā)部署，而機(jī)器人的操作和維修也需要人類技師。同樣，以人為核心的數(shù)據(jù)收集和標(biāo)簽分類工作也需要人類參與，以便為人工智能提供“燃料”，助力自動(dòng)化經(jīng)濟(jì)發(fā)展。

許多行業(yè)將根據(jù)“人類-數(shù)字共生”的嶄新模式獲得重塑。以教育領(lǐng)域?yàn)槔斯ぶ悄芸蓳?dān)任24小時(shí)在線的老師和教練，而人類老師則是智慧的導(dǎo)師和激勵(lì)者；在醫(yī)療領(lǐng)域，人工智能將協(xié)助人類醫(yī)生，充當(dāng)精準(zhǔn)的、有針對(duì)性的診斷引擎，而人類醫(yī)生則負(fù)責(zé)與患者溝通和做出最終的決定。

在向自動(dòng)化經(jīng)濟(jì)轉(zhuǎn)型的進(jìn)程中，政府和民間機(jī)構(gòu)需要攜手努力，共同試驗(yàn)和樹立最佳模式，幫助勞動(dòng)者做好準(zhǔn)備，并對(duì)面臨失業(yè)風(fēng)險(xiǎn)的人進(jìn)行再培訓(xùn)。中國(guó)較早采用自動(dòng)化技術(shù)，而且有望成功從一個(gè)貧困的農(nóng)業(yè)經(jīng)濟(jì)體轉(zhuǎn)型為一個(gè)工業(yè)強(qiáng)國(guó)，說明中國(guó)的實(shí)踐可以提供寶貴的經(jīng)驗(yàn)。

二戰(zhàn)后，美國(guó)的商業(yè)慣例成為了全球標(biāo)準(zhǔn)，世界各地的企業(yè)都立志與之看齊，在此次疫情的推動(dòng)下，中國(guó)將加快推進(jìn)自動(dòng)化、人工智能和機(jī)器人技術(shù)深入各行各業(yè)，同時(shí)為其他地區(qū)探索未來方向提供可借鑒的思路?！?/p>

The world after covid-19

Kai-Fu Lee on howcovid spurs China's great robotic leap forward

The pandemic is accelarating a shifttowards people-less companiesthat will eventually characterise business everywhere

DURING THE SARS crisis almost 20 years ago, shops in Beijing were completely closed. It forced one retailer, Jingdong Century Trading Co, to try its luck online.That retailer was JD.com, today one of China’s e-commerce giants. Fast forward to 2020. Amid the covid-19 crisis, JD.com managed a 20% spike in sales with the help of a new, automated warehouse that can process more than 1.5m orders a day. In Wuhan, it delivered packages using robots and drones. It plans to open 1,000 automated restaurants this year across China.

The efforts are representative of a broader shift amid the pandemic towards automation, artificial intelligence and digitisation. The Chinese economy is undergoing a great robotic leap forward, as it removes human touch-points—literally—in its operations. Online businesses, algorithms and automation save costs, boost efficiency and protect public health. Though the shift predates covid-19, the crisis has accelerated it. Social distancing via automation will have wide-ranging implications. As goes China, so may go business everywhere.

People used to expect to interact with other people to get things done. No longer. Much work can be largely handed over to a combination of software and robotics. That’s true for blue-collar jobs in health care, food service, delivery, manufacturing, logistics, transport and education. And it is increasingly happening for back-office white-collar jobs in finance, customer service, sales, human resources, law and accounting.

Historically, automation tends to happen when economic difficulties coincide with maturing technologies. Companies feel they need to cut costs by slashing jobs and trying out new technologies. And once a company has replaced an employee with a robot and proven its efficacy, it is unlikely to go back. Robots don’t get sick. They don’t strike. They don’t demand higher wages for dangerous jobs. In fact, they are ideal for dangerous jobs, which in a pandemic is any job that requires interaction with people. It is no wonder that David Autor, an economist at MIT, calls the covid-19 pandemic and economic crisis “an automation-forcing event”.

China is uniquely positioned to lead the world in the automation economy. Though the country has a large workforce, the cost of labour has increased ten-fold in the last 20 years and is now more than twice as high as Vietnam’s. As the workshop of the world, it has an incentive to automate its manufacturing sector, which enjoys a lead on high-quality products. China is now the world’s largest market for industrial robotics and the fastest-growing, surging by 21% to $5.4bn in 2018. This represents a third of global sales. As a result, Chinese companies are developing a leg-up on the world in terms of how to work with llic colleagues.

This has spilled over to domains beyond manufacturing. When the pandemic was spreading rapidly in Wuhan in February and the massive Huoshenshan Hospital was built in ten days, a fleet of robots was scurrying inside for disinfecting and delivering medical supplies. These machines are used across China in schools, hospitals and commercial buildings. Keenon, a robotics company in Shanghai, has developed an autonomous vehicle to disinfect areas, using a combination of LIDAR, machine vision and sensors.

Recently, when I was in quarantine at home in Beijing, all of my e-commerce packages and food were delivered by a robot in my apartment complex. The item would be placed on as turdy, wheeled creature resembling R2D2. It could wirelessly summon the elevator, navigate autonomously to my door and then call my phone to announce its arrival. I could then take the delivery and the bot would return to reception.

A food-delivery firm, Meituan, has introduced a “zero contact” service, where the meals are delivered to specific drop-off points, but the driver and customer need never interact. The company is testing self-driving delivery vehicles. WeChat, a popular social-media and payment platform, has developed a system for people to use their phone to read a restaurant menu, order a meal and settle the bill, with either a human or a robot delivering the food to the table. Robot servers today are both gimmicks and safety measures, but tomorrow they may be a normal part of table service for most restaurants, save for posh ones.

However, the impact of automation is likely to be higher for white-collar jobs in the short term. Although basic, routine tasks like assembly-line work are easy to automate, much manual labour is hard for robots. “Intelligent automation” in the 21st century will be different from the rudimentary physical automation of the 20th century, since today’s robots require drawing together mechanical engineering, AI for perception and fine-motor manipulation. Yet as white-collar employees work from home during the pandemic, everything happens online and all tasks are translated into data, it is a small step to have machines take over completely. Companies offering “robotic process automation” are experiencing a boom in sales since the crisis began.

I’ve seen these trends develop as a technology investor in China—and had a front-row seat during lockdown. Zhuiyi Technology, a company in our portfolio, develops software for call-centre automation. During the pandemic, the credit-card department of a large Chinese bank used the system to call its customers, managing 350,000 calls a day, or the equivalent of 1,200 human customer-service representatives. These conversational bots not only reduce cost, but also improve customer satisfaction and boost revenue. The company has since expanded its range to include AI telemarketers, AI analysts, AI trainers, AI assistants and soon.

We can already see the contours of the post-covid business environment taking shape. Everything that can be cost-effectively automated will be, removing people from the process not just for profit and performance but for health and safety. If robots and software were appealing before the pandemic, the reasons not to adopt the technologies are even thinner now, whether a vaccine against covid-19 is developed or not.

Although automation changes some jobs and decimates others, many new jobs are created too. For example, the new digital infrastructure of data centres, 5G equipment, and software will need human workers, as will the operation and repair of robots. So will human-centric data collection and labelling, as the “fuel” that powers AI and facilitates an automation economy.

Many sectors will be reimagined in the form of human-digital symbiosis. In education, for example, AI will become the tutor and always-on instructor, while the human is the wise mentor and motivator. In healthcare, AI will be the accurate, targeted diagnosis engine that assists the human doctor, who communicates with patients and makes final decisions.

The public and private sectors need to work together on the transition to the automation economy. They must experiment and establish the best practices to prepare the workforce and retrain those at risk of unemployment. China’s early adoption of automation technologies—and its successful transition from an impoverished agrarian economy to an industrial powerhouse—means its experience offers valuable lessons.

After the second world war, American business practices became the global standard to which all companies around the world aspired. Accelerated by the pandemic, China will be the economy that drives automation, AI and robotics deep into businesses and industries, while showing the way for others to follow. ■

Kai-Fu Lee, Ph.D., is the Chairman and CEO of Sinovation Ventures.

The Economist Newspaper Limited, London 25th June 2020

本文由李開復(fù)博士應(yīng)邀為《經(jīng)濟(jì)學(xué)人》撰文，原文：https://www.economist.com/by-invitation/2020/06/25/kai-fu-lee-on-how-covid-spurs-chinas-great-robotic-leap-forward。