Head of Supply Chain and Transport Industries, World Economic Forum
Christmas 2017 was a busy one for retailers. During peak season, Amazon reported records for holiday shipping as they handled deliveries to 185 countries. UPS, meanwhile, said it was breaking records on package returns, having processed more than a million daily in December. With a 23 percent global growth in the last year, e-commerce continues to post solid performances, but stress and confusion are routine for merchants, logistics companies and consumers. According to a UK survey, over half of Christmas parcels failed to turn up on time in the run-up to Christmas, and parents panicked as popular Christmas toys sold out at the UK’s biggest retailers.
Artificial intelligence can help bring order to this chaos. Here’s how:
On-time or faster deliveries through AI
Delivery delays and out-of-stock items cause consumer frustration. Wouldn’t it be ideal if companies knew who was likely to buy what and when? Supply chain professionals could then make sure they had the right manpower, warehouse and transport capacities, or even send goods to be stored close to delivery addresses before we buy them (a process known as “anticipatory shipping”). Amazon has been experimenting with AI for years – for example in its cashier-less store – but other companies are close behind. Otto, a German e-commerce merchant, has created a system that analyses around 3 billion transactions and 200 variables, including sales data, website searches and weather information. With 90 percent accuracy, it is able to predict sales behaviour in the next 30 days. Now, the company purchases around 200,000 items a month without human intervention. Surplus stock has declined by a fifth and product returns by more than two million items a year. While customers receive their orders more promptly and efficiently, the planet benefits too, as fewer shipped packages need to be sent back.
Consumers get what they wish for
A major challenge in the supply chain is predicting next season’s hit products. Merchants have to make choices early on, and accept the risk that their stockpiled goods may not be bought. On the other side, consumers also have to hope that retailers have in stock what they want to buy. Supply-demand uncertainties and mismatches are costly and inefficient for all participants. Brand reputation is at stake, and consumers pay a higher price to cover the risk. One AI-powered approach is dubbed optimised line planning, which integrates data such as internal sales and customer records, competitive intelligence, trend analysis and social media preferences to create a customer profile or persona. This customer segmentation allows retailers to determine the selection of products that will resonate best with each persona. Designers can create placeholders for next season, and even calculate expected revenues. This provides confidence that both sales targets and the needs of the consumer will be met.
Goods travel safely
When supply-chain professionals know what is happening, they can ensure packages will arrive despite any adverse conditions. The Tel Aviv maritime data provider Windward says it is able, through AI, to predict shipping safety worldwide. The Israeli startup signed a deal with London-based insurance market Lloyd’s in November 2017, with Windward providing Lloyd’s member companies software that forecasts maritime hostilities or accidents at sea. Flextronics International – which counts Apple, Microsoft and Ford Motor as customers – has pioneered software that generates real-time alerts of supply-chain disruptions throughout its 14,000-strong network of global suppliers. The AI-based system helps predict actual and potential problems, such as supplier delays, strikes, earthquakes or tsunamis, and allows the relevant teams to make informed decisions to keep inventory moving and consumers happy.
Fewer goods are stolen
Some orders don’t arrive due to crime. The FBI estimates that cargo theft costs US businesses more than $30 billion each year. Facial recognition and the detection of suspicious behaviour can help prevent this. At the Chinese AI research company Yitu Technology, the pass cards of staff and visitors taking the lifts to floors 23 and 25 are read automatically – no swipe required – and each passenger is deposited at their specified floor, and only there. Cameras record everyone entering the building, and track them once they are inside. “Our machines can very easily recognise you among at least 2 billion people in a matter of seconds,” says Yitu co-founder and chief executive Zhu Long. Yitu’s generic portrait platform already contains 1.8 billion photographs of those logged in the national database and everyone who visited China recently. 320 million of the photos have come from China’s borders, where pictures are taken of everyone who enters and leaves the country. This cutting-edge technology can be used to protect any kind of asset – from vehicles to shops, to warehouses and even entire cities. In Boston, intelligent security cameras are even anticipating crime. The security system monitors feeds in real time and alerts authorities the moment it identifies unusual activity.
Quicker and better customer service
According to a survey conducted by Genesys in 16 key economies, the cost of poor customer service is estimated at $338.5 billion per year. Consumers disappointed when drivers don’t show up or products are sold out can still be satisfied when they can act based on timely and correct information. AI tools can speedily provide proper information to customer service staff, and improve customer-agent conversations without replacing them (chatbots have been shown to thus far lack empathy and problem-solving capability). At the same time, digital voice-activated assistants could provide an opportunity for progress – just imagine you ask: “Contact DHL and find out when my order will arrive,” and your digital assistant connects to a chatbot, finds the information and gets it to you promptly. The assistant would soon even be able to use your voice, if required. A paper published by Google in December 2017 reveals details of a text-to-speech system called Tacotron 2, which claims to be able to replicate near-human accuracy when imitating a person speaking.
These developments don’t come without risks. Concerns about job losses and out-of-control machines that have the potential to teach themselves are daily topics in the news. As AI-enabled robots have started to work alongside humans as autonomous vehicles, security cameras and digital assistants, some experts have said they should be fitted with an “ethical black box” to keep track of their reasoning, and enable them to explain their actions when accidents happen – such as the fatal self-driving car crash of a Tesla Model S in May 2016. Others think the behaviour of robots should be viewed similarly to that of humans, which often remains a mystery. While there are many voices urging caution, there are also positive ones. Stephane Kasriel, the CEO of Upwork, predicts “that there will not be a shortage of jobs in the future, but rather a shortage of skills to fill the jobs“. History has taught us that, over time, concerns with respect to new technologies will be addressed so that consumers can reap the benefits.
AI is on the rise and is sure to enter into many parts of our lives. Considering the impact it can have on consumer satisfaction, it would not be surprising if consumers not only embraced the intelligent supply chain but demanded more AI solutions in the purchasing process. Meanwhile, investors soon may no longer fund businesses that aren’t planning to incorporate AI in some form.
Wolfgang Lehmacher, Head of Supply Chain and Transport Industries, World Economic Forum