工作岗位将被人工智能取代，如何应对？

自动化正在通过各种方式逐渐渗透至职场，也让雇员们开始担心，科技会对自身工作带来哪些变化，甚至是否会取而代之。牛津经济研究所在2019年6月发布的一篇报道预测，到2030年，全球8.5%的制造业岗位，约2000万份工作，将被机器人取代。

位于匹兹堡的卡耐基梅隆大学计算机科学学院的教授兼临时院长汤姆·米切尔表示，从这个角度来思考自动化和工作之间的关系是错误的。相反，人们应该审视自己工作所涉及的任务，并评估一下让这些任务实现自动化的难易程度。

他说：“一些人从事的是单一任务工作，例如收费站[操作人员]。这些人会遇到麻烦，因为他们的工作将成为自动化的对象。”当然，这对于他们来说是个坏消息，但对其他人而言又意味着什么呢？

面临风险的任务

一些任务评估起来并不容易。2013年的调查《就业的未来：计算机化对工作的影响》发现，约47%的工作都会因为人工智能的进步而面临高风险。

该研究的合著者、《科技陷阱：自动化时代的资本、劳动力和电力》（The Technology Trap: Capital, Labor, and Power in the Age of Automation）一书的作者卡尔·本尼迪克特·弗雷博士称，有关自动化影响的预测呈现出两极分化的趋势：一派人认为机器人将取代诸多工作，并让很多人失业，而另一派则认为它将改变工作的性质。

他补充说：“但这也意味着大量的人可能会失业，因为随着工作性质的改变，他们的技能成为了一种过剩的事物。”

这并非是什么带有未来主义色彩的假想。麦肯锡全球研究所（MGI）的合伙人迈克尔·楚伊博士表示，几乎现有工作中的半数任务在理论上都可以通过科技手段来完成。

而且工作发生改变的不仅仅是低收入工作者。楚伊和他的团队预测，约60%的工作都含有30%以上可以被自动化的任务。首席执行官、金融顾问、保险代理和其他一些工作都是如此。

无论职级高低，那些工作将因为科技而发生彻底变化的员工对于自动化的含义漠不关心。即便科技不会让他们完全失业，但他们仍然有必要了解自己的工作会发生哪些变化，以及何时发生变化。

预测变化率

尽管我们很难精确地给出自动化取代员工工作的具体时间表，但有一些明确的指标会告诉人们接下来会发生什么，同时，我们还面临着可能会拖慢这一流程的障碍。

在首批指标中，其中一个便是构成自身工作的任务类型。MGI发现，可预测的体力劳动、数据处理和数据自动化均属于非常容易受到自动化影响的工作。但他们的研究还显示，人们很难为其他职务开发有效的技术解决方案，例如不可预测的体力劳动、与股东的互动、专长运用以及人员管理等。因此，尽管聊天机器人可以回答一些基本的问题，机器人可以从仓库中拿出物件进行打包，但我们有理由相信，建造工作、林业工作或室外动物养殖在短期之内不会面临任何风险。

楚伊表示，自动化的发展和采纳过程较为缓慢，然而它在成为主流之后便会迅猛发展。他说：“对于过去几十年中的任何技术来说，以实际中存在的积极商业案例为例，商业可用性距离该技术在经济中的采用达到稳定时所需的时间大约为10-30年。我们的模型给出的时间是8-28年。”

有时候，你甚至可能会成为这个流程的一部分。埃森哲正在让自家员工与其客户寻找那些能够被自动化的关键任务。例如，埃森哲的运营部门在其田纳西诺克斯维尔中心举行黑客马拉松比赛。通常在周五下班之后，雇员们可以留下来享用比萨晚餐，然后与人工智能专家和数据科学家会面，讨论如何解决自动化解决方案中的问题，每月举行一次。

埃森哲运营部门的首席执行官德比·珀里舒克称，雇员们对于其工作的自动化并不感到害怕，因为领导层鼓励他们通过各种方式来创造更多价值，让工作更加有趣。此举也打消了雇员对自己工作将被取代的顾虑。她说：“是否会有那么一天，任何事情都已经100%的自动化，无需监控，无需进行额外的培训？在我看来这一天是不存在的，要让工作中的每个单一流程都实现自动化是不现实的。”

珀里舒克的团队还与客户展开合作，使用技术工具来实现工作流程的自动化。埃森哲的雇员负责开发监控客户雇员工作的人工智能工具。珀里舒克表示，整个流程是透明的，许可由客户授予，雇员也知道自己处于被监控的状态。埃森哲的人工智能和技术专家发现了可实现自动化的领域，并在这一过程中征求雇员的意见。雇员甚至可以参与帮助“培训”实现其工作自动化的人工智能技术，绘制流程并制定结果标准。

自动化道路上的障碍

楚伊指出，除了技术开发以及培训其如何正确工作所面临的挑战之外，人工智能的广泛采用还面临着一些典型的障碍。成本是其中一个。他表示，即使流程得以确认，而且也存在自动化的商业案例，但大多数新技术都有着相对较高的价格。

楚伊说：“人们对美国200万的卡车司机感到担忧。”即便这一技术做好了部署的准备，而且存在积极的商业案例，但他估计，要取代美国所有的卡车，其成本可能会达到数千亿美元。这类成本是一个很大的障碍。

其次，人们对于技术变革存在一些善意、传统的抵触观念。当然，无人驾驶卡车可能在很多方面是一个不错的解决方案，然而，对于在路上以70英里/时的速度行驶的无人驾驶交通工具，并非所有人都能够接受。

米切尔提到了在快餐店实现众多任务自动化的现有技术。他说：“事实在于，我依然喜欢与人打交道。我并不觉得人们意识到，自己在任何一家零售店面支付货款时有多大一部分价值应该归属于这种人际互动。客户对于与全自动化的机器打交道会有多大的抵触情绪？”

他还指出，很多领域在很长一段时间内将采用部分自动化。因此，关注自身所在行业的发展以及已开发技术的类型，对于预测工作如何变化以及何时变化至关重要。（财富中文网）

译者：冯丰

审校：夏林

Automation is increasingly making its way into the workplace, raising concerns among employees about the ways technology will change their jobs—or eliminate them entirely. A June 2019 report by Oxford Economics predicts that 8.5% of the world’s manufacturing positions alone—some 20 million jobs—will be displaced by robots by 2030.

But that’s the wrong way to think about automation and jobs, says Tom Mitchell, professor and interim dean of Pittsburgh-based Carnegie Mellon University’s School of Computer Science. Instead, you should look at the tasks involved in your job and evaluate how easily those tasks can be automated.

“Some people have a single task job, like toll booth [operators],” he says. "Those people are in trouble because their job is going to be automated.” That’s bad news for them, of course, but what does it mean for you?

Tasks that are at risk

Some tasks aren’t easy to evaluate. A 2013 paper, “The Future of Employment: How Susceptible are Jobs to Computerisation?” found that roughly 47% of jobs were at high risk of being automated with advances in artificial intelligence.

Carl Benedikt Frey, Ph.D., co-author of that paper and author of The Technology Trap: Capital, Labor, and Power in the Age of Automation says predictions around automation’s impact have become very polarized: Either you believe that the robots are coming for many jobs—leaving many with no employment—or you believe it’s going to change the nature of work.

“But that also means that lots of people are probably going to lose their jobs because their skillsets are becoming redundant even as the nature of work changes,” he adds.

This isn’t some futuristic hypothetical. Michael Chui, Ph.D., a partner at McKinsey Global Institute (MGI), says roughly half of the tasks people perform at work could theoretically be done by technology that exists today.

And it’s not just low-income workers whose jobs will change. Chui and his team estimate that roughly six out of 10 jobs are made up of 30% or more tasks that can be automated. CEOs, financial advisors, insurance agents, and others all fall into this category.

Regardless of their title, those whose jobs will be transformed by technology care little about the semantics of automation. Even if technology won’t leave them entirely unemployed, they still need to keep abreast of how their jobs will change—and when.

Predicting the rate of change

While it’s difficult to accurately pinpoint a specific window of when automation will encroach workers’ jobs, there are some good indicators of what’s coming, as well as some obstacles that can slow down the process.

One of the first indicators is the type of tasks that make up your job. MGI finds that predictable physical work, data processing, and data automation are all highly susceptible to automation. But their research also shows it’s tougher to find effective technology solutions for other roles, like unpredictable physical work, interactions with stakeholders, applying expertise, and managing others. So, while chatbots may be able to answer basic questions, and robots may be able to pick items out of a warehouse for packing, it's safe to assume that construction, forestry work, or raising outdoor animals likely aren’t at risk any time soon.

Chui says that automation develops and is adopted slowly, but comes on fast once it’s hit the mainstream. “For any technology in the past few decades, the time between commercial availability—say that there’s an actual positive business case—and the plateau in adoption of this technology across the economy, is roughly one to three decades,” he says. “We model it out as eight to 28 years.”

Sometimes, you may be even be part of the process. Accenture involves its own employees and those of its clients in identifying key tasks to be automated. For example, Accenture’s operations department holds hack-a-thons at its Knoxville, Tennessee center. Once a month, usually after work hours on a Friday, employees can stay for a pizza dinner and meet with AI specialists and data scientists to figure out how to address problems with automation solutions.

According to Debbie Polishook, group chief executive at Accenture Operations, employees aren’t afraid to automate parts of their jobs because of leadership encouraging them to find ways to add more value and do more interesting work. This alleviates fears that their roles will be eliminated. “Do I see a day when everything is 100% automated with no supervision, no additional training required? I really don’t see a day where that's true for every single process in the workplace,” she says.

Polishook’s team also works with clients, using technology tools to automate work processes. Accenture’s employees oversee A.I.-powered tools that monitor how clients’ employees work. The process is transparent—permission is granted by the client and employees know they’re being monitored, Polishook says. Accenture’s A.I. and technology specialists identify areas that could be automated, consulting employees along the way. Employees may even be enlisted to help “train” the A.I. that will automate their work, mapping processes and setting outcome standards.

Barriers to automation

Beyond the challenges of developing and training the technology to work properly, a number of barriers typically stand in the way of widespread adoption, Chui says. Cost is one. Even when the process has been identified and there’s a business case for automating it, most new technology has a relatively high price tag, he says.

“People worry about two million truck drivers in the U.S.,” Chui says. Even if the technology was ready to deploy and there was a positive business case for them, he estimates it would cost hundreds of billions of dollars to replace every truck in the U.S. That kind of cost is a big barrier.

Then, there’s good, old-fashioned resistance to technological change. Sure, autonomous trucks may be a great solution in many ways. But not everyone is comfortable with the thought of vehicles without drivers barreling down the road at 70 miles per hour.

Mitchell points to the existence of technology to automate many tasks in fast food restaurants. “The truth is, I still like interacting with humans,” he says. “I don’t think we know how much of the value that we’re paying for in any given, say, retail outlet, [is in the interaction]. How much customer resistance would there be to dealing with full automation?”

He adds that there will be a long path of partial automation in many sectors. So, paying attention to developments in your sector and the types of technology being developed is essential to predicting how and when your job will change.