Artificial Intelligence


A New Reality for Shipping

Companies that fail to take advantage of the benefits of Artificial Intelligence (AI) will be at a commercial disadvantage. That’s the message ringing loud and clear from academics and industry professionals alike. Put simply, AI technology is a game changer – in our business and personal lives.

Take a recent high-level research report by MITSloan, in collaboration with The Boston Consulting Group (BCG), entitled ‘Reshaping Business with Artificial Intelligence’. It surveyed 3,000 managers, executives and analysts in many industries across the globe. Research indicates that around 85 percent of respondents believe AI could have dramatic commercial benefits.

However, only about one-fifth of companies had incorporated AI in any form into their processes or service offerings. A key conclusion of the report is that companies must act now to plan for AI, and those that fail to do so will have the “playing field tilted evermore steeply against them”.

From the outset it is worth reminding ourselves of a definition of AI from the Oxford Dictionary, “AI is the theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision making, and translation between languages.”

According to Professor Yi Yang, Assistant Professor, Information Systems, Business Statistics and Operations Management, Hong Kong University of Science and Technology, there are four main aspects to the successful development of AI.

He stressed, “Massive amount of valuable data, fast computing power, efficient and effective algorithms and vast financial support from private and public sectors are responsible for the fast development of deep learning as well as AI.”


By 2020, the global demand for AI-related products is expected to reach US$47 billion, with healthcare and financial services industry sectors continuing to drive its uptake and evolution, as reported by Mondaq.

Indeed, AI stands at a high point in its evolution, according to Dr Yvo Saanen, Commercial Director and founder of TBA Group. He believes the combination of factors cited above makes the power of learning behaviour and pattern recognition software much greater. This has implications for industries that want to harness this technology.

He stressed, “The container industry is very repetitive. Hence, learning from the past helps improve decision making in the future. So, within the terminal operating systems, or within surrounding intelligence modules, there’s great potential to use AI technology to address these issues. One hurdle to overcome is quality of data; today the quality is still quite poor,” said Saanen.

“So, until we get solid information throughout the supply chain, shared and put into algorithms, it will be very difficult to make good high-quality decisions,” he added.

In the longer term, Saneen expects to see AI incorporated inside automated equipment and vehicles, to make them more autonomous. He also foresees opportunities for automatically generated stacking systems to operate cranes and move containers, and the co-existence of partial autonomy and central optimisation.

The Port of Rotterdam is teaming up with IBM to work on applications of the Internet of Things (IoT) and AI. A recent initiative is to use sensors to gather multiple datastreams to enable port authorities to predict the best time (based on water level) to have ships arrive and depart Rotterdam.

Another innovation incorporating AI is DHL’s Global Trade Barometer, launched earlier this year, which derives predictions for global trade by evaluating large amounts of logistics data with the help of AI.

Tim Scharwath, CEO of DHL Global Forwarding & Freight, said, “The DHL Global Trade Barometer shows impressively how digitalisation – with the use of Big Data and Predictive Analytics – opens up entirely new opportunities that we can use for the benefit of our customers.”


At a recent seminar held at Hong Kong Science Park, with the theme ‘AI Platform for Smart Applications’, one of the key considerations from a variety of high level speakers was the essential requirement of data for effective AI – as the saying goes ‘garbage in, garbage out’. Thus, companies need to have built robust information infrastructure, so algorithms can learn effectively by monitoring the data, said Yang.

Professor Yi Yang concurs that AI is only as good as the data provided. Yang stressed that companies need to focus on getting valuable data. It is a question of quality over quantity.

Yang added, “Before blindly collecting data, a company needs to decide what its goal is. Reduce shipping cost? Optimise transit routes? Once the goal is set up, then the company needs to figure out what types of data to collect, from traditional structured accounting data to non-traditional unstructured data generated from sensors, GPS devices, and so on. ”

However recent developments in transfer learning, a machine-learning technique that focuses on storing knowledge gained while solving one problem and applying it to a different (but related) problem, has started to draw attention from both academics and industry thinkers.

Yang explained, “For example, an AI model trained to recognise cars could be reused to recognise trucks. Transfer learning can significantly reduce the effort of obtaining data and training an algorithm model. Therefore, we expect to see successful AI applications where data is limited or even not provided.”

Yang supports recent analysis by a wide variety of commentators that AI can positively impact every aspect of supply chain management.

He said, “The self-driving fleets can reduce, estimated to be around 80 percent, shipping cost, which benefits companies in a variety of sectors that rely on the vehicles for shipping.”


Another area will be warehouse management, according to Yang. The core of successful supply chain management is efficient warehouse and inventory management. By providing accurate demand forecasting, he added, AI technology will reshape warehouse and inventory management process.

A Financial Times report recently investigated the development of robots in ecommerce warehouses. The reporter visited Ocado’s warehouse facility near Andover, in the south of England. A centralised computer system using a planning algorithm communicates with robots deployed within the vast warehouse. These robots can collaborate in a swarm and, remarkably, can complete a 50-item customer order within minutes.

At a separate DHL facility, some collaborative robots (aka cobots) are being installed to work alongside humans, and these cobots can be shown tasks and respond to touch without the need for preinstalled programming, supporting repetitive and physically demanding tasks in logistics operations.

Some well-informed commentators have speculated that the implications of biometric facial recognition technology could eliminate the need for scanners, it could guide vehicles within and away from the warehouse. It could also improve safety standards and productivity levels as scanners would be surplus to requirements: someone would just need to glance at the products. According to DHL trials of so-called ‘smart glasses’ during order picking in logistics, which enable intelligent hands-free operations and harness Augmented Reality (AR), a 25 percent upturn in efficiency was observed.

Industry sources suggest characteristics of AI technology are now being used on mobile apps, where users can photograph a desired item and visual recognition technology can search for a matching item.


According to Yang, it is Deep Reinforcement Learning, an emerging AI technique, which will significantly enhance operations. Some commentators compare Deep Learning to the workings of the human brain that can sift through data more effectively and speeds up the analytical process. This technology is being incorporated in manufacturing operations in a joint venture including, Hitachi, Fanuc, and an AI start up called Preferred Networks Inc., as reported by Nikkei Asian Review. This should dramatically improve productivity.

Yang explained, “Unlike traditional supervised machine learning where domain knowledge has to be provided, Deep Reinforcement Learning can figure out how to do things that no expert need to or could teach them.”

“Take Chatbot for example, the Deep Reinforcement Learning powered Chatbot can observe from the conversations between human agents and customers, simulate the conversations with itself, understand what contributes to the success of a dialogue and eventually generate meaningful and accurate conversations with customers, which reduces operational costs,” he stressed.

Peter Spellman, Chief Technology Officer at INTTRA, a global technology platform enterprise, is confident Chatbots will become more ubiquitous in the coming years and offers great benefits to customers.

“Chatbot-AI application would recognise that a specific customer regularly books certain routes, learning from the customer’s past bookings and behaviours. The Chatbot might complete the booking, or it will know to escalate and engage customer support when exceptions are required,” said Spellman.

This will allow organisations to improve the booking process and free up staff to focus on more rewarding tasks. AI will help remove inefficiencies in the system by directing technology at areas that can be easily standardised, and which can benefit from processing power over a wide data set. This includes everything from prefilling of various forms, or aspects of customer service that will help customers find information and sources before they speak with a client.


For Spellman, this aspect to AI, the altering of working patterns is one of its key positives.

He explained, “AI is one of many in a series of disruptions over time, all of which change the shape of the workforce. AI programmes will continue to advance, but their shortcomings are why humans will remain key to AI implementation. Humans will handle exceptions and higher-end tasks, ones that require creative and original thinking, as well as leadership roles and other tasks.”

Spellman concedes AI will cause disruption in the process, as is the case for practically all new technologies. He cited a report by Gartner which suggests AI will create 2.3 million new jobs and “has the potential to enrich careers, reimagine old tasks and create new industries.”

Spellman foresees information technology and AI as differentiators, and urges companies not to be on the wrong side of the digital divide. INTTRA’s 2032 White Papers outlines where the logistics sector may be heading.

He echoes the sentiments of academics and other industry thinkers about the ability to store unlimited data in a cloud environment can begin the process of reasoning with it to create AI applications. This should enable the development of more meaningful AI applications.

In the future, AI should increasingly help with other critical tasks while working in conjunction with humans. For example, an error in Harmonised System (HS) codes, which is not uncommon, but necessary for shipments, can cause significant delays and increase costs. AI coupled with Machine Learning could correct such errors without humans, unless it’s unable to correct the issue on its own, according to Spellman.

Companies are looking at digitalisation, AI and all forms of technology to reduce costs through innovation, improve efficiencies and to generate higher revenues. Around 800,000 container orders are processed over INTTRA’s platform weekly through more than 60 carriers. The intersection of AI with blockchain, quantum computing and the IoT and other transformative technologies is a key future trend.

Spellmen said, “AI promises to further expand on the advantages of digitalisation, which is why companies are increasingly looking at what it can provide.”


According to Professor Yang, adopting AI technology requires a holistic solution from data to IT infrastructure to manpower to organisational structure. By automating processes and making better decisions, early adopters of AI will enhance productivity and gain obvious competitive advantage, he claims.

“The AI system will also generate valuable feedback and new operational data which can further tune up the system setting. All those advantages will become barriers for competitors,” said Yang.

In a recent commentary within the Council of Supply Chain Management Professionals (CSCMP) Supply Chain Quarterly, the author warned supply chain companies that they need to fully understand the impacts of a solution before adopting new technology, such as AI. For instance, operational costs may be reduced. But does this come at the expense of customer service quality and will this compromise customer loyalty?

For Professor Yang there are two main challenges for the development of AI. The first is a lack of ‘multitask AI’ or ‘generalised AI’. Yang said almost all AI technologies are highly specialised with no AI system that can be trained to drive a car, play golf and translate languages.

“The second challenge is the lack of AI manpower. More universities are providing courses focusing on the skills needed for AI development – computer science, statistics, etc. But there are still not enough people to enable every business and organisation to take up AI technology,” said Professor Yang.

He also warned that AI is still decades away from true human-level intelligence and therefore AI practitioners from industry and academia should not overhype what AI can do.

He concluded, “Overpromising would result in investment loss as well as disappointment and distrust in AI capabilities.”

Ultimately, as AI becomes more entrenched in our society, its implementation will raise many ethical and commercial challenges, but there can be little doubt the AI revolution bandwagon is set to roll on and on.