超市的自动结账机真的会提高效率么

At a conference recently, I heard the chief executive of a supermarket chain proudly claim that installing automatic checkout machines was improving the company’s productivity.

在最近的一个会议上，我听到一家连锁超市的首席执行官自豪地宣称，安装自动结账机提高了公司的生产率。

This is wrong, of course. These machines have replaced traditional checkout equipment — and their paid human operators — with unpaid customer labour in scanning and bagging. The company’s measured revenue per employee hour will certainly be rising. True productivity will not. Economic welfare in the round is reduced by the struggle we customers have with the still-imperfect new machines.

这个说法当然是错误的。这些机器用顾客扫描和装袋的无偿劳动取代了传统的结账设备——以及它们花钱聘请的收银员。如果计算该公司员工每小时为公司带来的收入，这个数字肯定会增加。实际的生产率却不会上升。整个社会的经济福祉还会因我们顾客要费力操作尚不完善的新机器而下降。

This raises the question of what would count as a genuine productivity gain. The retail sector made massive strides in productivity during the 1990s, mainly thanks to the use of the new information and communication technologies in logistics. Perhaps automation has further to go in terms of delivering products to the supermarkets. But what would higher productivity look like in terms of actually getting groceries into customers’ shopping bags?

这引出了如下问题：什么才算真正的生产率增长？上世纪90年代，零售业在生产率方面取得了重大进步，这主要得益于物流领域使用了新的信息和通信技术。或许自动化在将商品运送至超市方面还有进一步发展的空间。但是就将食品杂货放进顾客的购物袋的实际过程而言，更高的生产率会是什么样子的？

The answer might well lie with Amazon’s experimental store, Amazon Go. Customers put what they want in their bags and walk through a turnstile. A proliferation of cameras — and an algorithm — watch them, add up their bills and charge it to them. There is no checkout process at all.

答案很可能蕴含在亚马逊(Amazon)的实验性门店Amazon Go中。顾客把他们想要的东西放在袋子里，然后走过闸机。门店里大量时刻看着顾客的摄像头和算法帮助顾客结算账单。完全没有结帐过程。

This is spot on. In many retail industries, higher productivity means faster service. There are many routine services where getting more for less requires less time to be spent performing them. This applies to parts of many sectors of the economy. There are past examples — think of the impact of the washing machine on doing the laundry or the ATM on taking money out of the bank — but now we are seeing much more automation in new areas such as legal search, scrutiny of medical tests and buying train tickets online. There is surely much further to go as artificial intelligence advances.

这正是我们想要的答案。在许多零售行业，生产率越高意味着服务越快。在许多常规服务中，以更少成本获得更多就要求缩短这些服务所花费的时间。这适用于经济中的许多领域。过去有过这样的例子——想想洗衣机对洗衣这件事、或者ATM对取钱这件事的影响——但现在我们在法律查询、对医学测试的审查和在线购买火车票等新的领域看到了更多的自动化。随着人工智能技术的进步，肯定还会有进一步的发展。

In other types of service, however, it is not productivity that matters, but rather quality. The economically better outcome is likely to involve spending more time, not less, on the delivery of these non-routine services. Examples might include caring for a very sick patient in hospital, or preparing a special meal. Many existing sectors will include both activities we want to see done faster and those which would be improved if they happened more slowly. In either case, time is the right productivity metric.

然而，在其他类型的服务中，重要的不是生产率、而是质量。要获得经济上更好的结果，可能需要花费更多（而非更少）的时间来提供这些非常规服务。这方面的例子可能包括照料一位病重的住院病人，或者准备一顿特殊的饭。在现有的许多行业中，都会既有我们希望缩短完成时间的活动，又有花费更多时间会更好的活动。无论哪一种情况，时间都是恰当的生产率指标。

The amount of money earned per hour spent on the service is going to become a decreasingly useful metric, at least for economists if not for shareholders. For there is an overlapping question about whether or not the services are provided in the market — by paid labour, or in the household — by our own, unpaid labour.

一项服务每小时所赚取的收入这个指标将日渐失去效用，至少对于经济学家来说如此，甚至对股东也是一样。因为有一个同时存在的问题，那就是这些服务是不是通过市场提供的，也就是说，这些服务是通过有偿劳动提供的，还是在家里、由我们自己的无酬劳动提供的。

Technology is also shifting this conventional boundary between what is considered to be part of the economy and what is not. As well as high street retailers foisting some of their work on to us, certain formerly intermediated activities are now done online at home, such as banking or booking a holiday. The time might come when clever algorithms are domesticated enough to enable us to do much more of the routine type of service activity for ourselves. Indeed, much of professional services from management consultancy to legal advice involve decision processes that can easily be codified.

技术也正在改变以下两者之间的传统界限：被视为经济组成部分的东西和不被视为经济组成部分的东西。不只是实体零售商正在将他们的一些活儿交给我们自己来完成，一些以前需要中介的活动现在也可以在家通过网络完成，例如银行业务或预定度假产品。有朝一日，聪明的算法可能会被训练得足以让我们能够自己完成许多常规的服务活动。事实上，从管理咨询到法律咨询的许多专业服务都包含可以轻易程式化的决策环节。

Businesses should apply this routine/fast and non-routine/slow lens to their activities to think about their productivity and, more importantly, how well they are serving their customers. Amazon has invested on a grand scale, and it shows: the company delivers a superb customer experience.

企业应该从这种区分常规（越快越好）与非常规（时间越长越好）的角度来审视它们的各项活动，以这种方式思考它们的生产率，而且更重要的是，它们提供的服务在多大程度上满足了顾客需求。亚马逊已经大举投资，并且这种投资产生了效果：该公司提供了极好的客户体验。

If the Amazon Go technology works, will harassed shoppers prefer to scan goods at a machine that constantly harangues them about getting the bagging wrong — or walk out of a store with no checkout at all?

如果Amazon Go技术成功了，那么烦躁的购物者是愿意在一台总是告诉他们装袋出错的机器上扫描货物，还是无须结帐直接走出商店呢？