你可以与你的供应链交谈吗？日期   / 2018-11-02

本文由美国麻省理工学院交通与物流中心正式授权联合通商科技中译刊登在供应链中，通过语音功能的加入，带有语音功能的数字设备（如亚马逊的Alexa和Google Home）的销售额预计将从2017年的20亿美元猛增至2022年的400亿美元。这种快速增长的新技术将会影响供应链发生许多变化，而远远超出了单纯通过语音命令下订单的这种功能。例如，在一些零售商店中，员工甚至客户可以通过beacon激活麦克风以重新订购缺货商品，他们所需要作的只是大声说出指令，就这么简单。这种功能为生产商提供有价值的洞察，并帮助他们了解并调整供应，来满足需求的实际情况。

此外，客户能够在商店中标示缺失物品，这样有助于解决库存问题，由于零售库存系统有时候会错误地显示货架上实际上缺少的物品。这对零售商来说是一个主要问题——因为它会导致销售额的下降。
除此之外，会话式商务也正在改变消费者与在线零售商的互动方式，从而改变其运营的供应链模式。例如，关于采购状态的例行更新现在也是会话商务的一部分。买家可以询问语音助理的订单信息，如交货时间。
未来，以消费者为中心的应用程序并不遥远。将物联网传感器连接到语音设备上，以提供产品推荐或特殊优惠，并收集有关产品需求 的有价值的信息。想象一下，当一位消费者通过与数字助手讨论她该买什么外套好。她的智能手机上的照相机传达她穿着的夹克的图像，该图像与其他相关数据结合起来，使零售商能够提出产品推荐，并改进服装的需求预测。
语音技术也可以实现实时交通工具的重新布线。这已经可以实现，这也是供应链中的重要的里程碑，例如完成装载或交付货物后便可以口头传达给设备，使信息完成实时更新。
执行上述任务所需的大部分技术都已到位。例如，谷歌翻译可以将实时交通标志转换成不同的语言，或者苹果iPhone X上的高分辨率摄像头，即使在用不稳定的手拍摄照片时，它也能记录非常详细的图像和4k视频。随着技术和业务流程创新的进展，在未来的应用中将有更多的可能性。
未来的系统可以通过分析语音模式来衡量顾客的情绪和购买倾向。这些数据将丰富对产品需求模式和需求预测的分析。
但是有些问题也需要进一步研究。例如，语音识别远非完美，自然语言处理仍然为人工智能解决方案造成太多含糊不清的概念。机器难以确定〝便宜〞等模糊含义的含义，而且他们在调整背景噪音方面并不如人类好，这在诸如工厂车间或交通繁忙的道路等环境中会存在问题。
尽管如此，随着会话式商务融入公司运营中，它有可能使供应链更贴近客户，员工和高级管理层。
本文基于麻省理工学院自动识别实验室主任Brian Subirana，麻省理工学院开放学习副校长Sanjay Sarma，麻省理工学院交通运输与物流中心副主任Jim Rice以及麻省理工学院Ken Cottrill编写的文章，全球通信顾问在麻省理工学院运输与物流中心，并在斯隆管理评论的Frontiers博客网站上发表。

Can You Talk to Your Supply Chain?

Sales through voice-enabled digital assistants such as Amazon’s Alexa and Google Home are projected to skyrocket from $2 billion in 2017 to $40 billion in 2022.  How will this fast-growing technology affect supply chains?

Many changes will happen with implications that go far beyond the convenience of placing orders by voice commands.

Some of the changes will happen at the retail store level. For example, simple beacon-activated microphones will allow retail store employees and even customers to re-order missing products just by speaking out loud. Capturing these demand signals earlier than highly structured and cyclical (daily, weekly) ordering systems, gives producers valuable insight into demand profiles to which they can tailor their supply operations.

In addition, enabling customers to signal missing items in the store could help to solve the Phantom Inventory problem, where retail inventory systems mistakenly show that items are available on the shelf that are, in fact, missing. This a major issue for retailers since it leads to lost sales.

Outside of the store, conversational commerce is changing the way consumers interact with online retailers and hence the supply chains that support their operations.

The rapid growth in audio digital assistants could

 change the way customers interact with supply chains

Routine updates on the status of purchases are now part of conversational commerce. Buyers can interrogate voice assistants for order information such as delivery times.

A more futuristic consumer-centric application – but one that may not be far away – is wiring voice devices with Internet of Things sensors to deliver product recommendations or special offers and collect valuable information on product demand. Imagine a consumer conversing with a digital assistant about a jacket she wants to buy. The camera on her smartphone conveys images of the jacket she is wearing. The image combined with other relevant data enables the retailer to make product recommendations and to refine demand forecasts for the garment.

Voice technology can enable real time truck/conveyance re-routing. This is already possible, even if only rudimentarily, with a simple Alexa app on the new Garmin Speak device. Important milestones in the supply chain such as the completion of loading or delivering a shipment could be communicated verbally into devices, providing a richer description of these milestones.

Much of the technology required to perform the tasks described is in place. For example, Google translate which converts real-time traffic signs into different languages or the high-resolution camera on Apple’s iPhone X that records extremely detailed images and 4k videos, even when pictures are taken with an unsteady hand.

More possibilities will become available as the technology and business process innovation advance.

Future systems could gauge customers’ moods and propensity to buy by analyzing voice patterns. Such data would enrich analyses of product demand patterns and demand forecasts.

There are issues that need to be researched further. Speech recognition is far from perfect and natural language processing still creates too many ambiguities for artificial intelligence solutions. It’s difficult for machines to pin down the meaning of indistinct terms such as “cheap,” and they’re not as good as humans at tuning out background noise, a problem in settings such as a factory floor or a heavily trafficked road.

Still, as conversational commerce becomes embedded into company operations, it has the potential to bring supply chains closer to customers, employees and senior management.

This post is based on an article written by Brian Subirana, Director of the MIT Auto-ID Laboratory, Sanjay Sarma, Vice President of Open Learning at MIT, Jim Rice, Deputy Director of the MIT Center for Transportation & Logistics, and Ken Cottrill, Global Communications Consultant at the MIT Center for Transportation & Logistics, and published in Sloan Management Review’s Frontiers blog site.