Artificial intelligence gives businesses a competitive edge

Despite increasing digitalisation in logistics, companies are still struggling with the transition to completely data-driven operations. It is still proving a challenge to collect relevant and accurate data from the many different partners and to act on it – especially in the field of logistics, where data loses value fast. This is because a truly data-oriented approach requires analytical skills, complex programming, integration, and interfaces that provide the users with effective decision-making support.

Limits of process optimisation
By contrast, artificial intelligence needs congruent and robust process programming to start with, but then uses the entire operational backend to consolidate data and draw conclusions from it. For example, AI-based software can analyse real-time data for each container in transit and process data regarding the country of origin and the forwarder to create a risk profile for each container and decide whether and where it should be checked.
AI systems can also help to optimise and execute processes. Logistics systems for transport management, dispatch planning and shipping are currently designed to maximise throughput, reduce costs and increase transparency. But process optimisation also has its limits, especially when it comes to managing exceptions. Dealing with exceptions in complex networks requires numerous system adjustments, which are costly and can overwhelm system users.

Just a tiny labelling error can lead to significant costs, delays and safety risks in the supply chain and require human intervention. With the help of AI-based technology, however, exceptions that occur in conjunction with less ambitious system adjustments can be dealt with more effectively. Companies that ignore the opportunities offered by AI for process optimisation will undoubtedly struggle to keep up.

In addition to making decisions based on extensive data sets, AI is also able to optimise success criteria. This optimisation extends beyond logistics costs to non-monetary parameters like the modal split ratio, resource utilisation, service level agreements, emissions and many more. AI can develop itself to balance out multi-dimensional optimisation targets and adapt quickly to changed circumstances or events when necessary.

An example of this is the way AI takes into account potential deviations in lead times as part of multi-echelon inventory optimisation. It is able to analyse numerous different elements and identify correlations that are too complex or too abstract for humans to fully grasp. In the very near future, it will be AI that is best placed to handle the ever-growing requirements influencing daily logistics decisions. This shift is likely to make logistics companies and the industry as a whole significantly more resilient. Thus, it is hardly surprising that AI is currently the subject of much debate. This new and disruptive technology does, however, come with its own set of challenges – such as the risks posed by unchecked AI: Artificial intelligence needs a high degree of freedom to work best, yet it also needs to be controlled to avoid unintended consequences when it comes to things like ethical considerations or the need to quickly regain full control if a recovery process is initiated. Governments and regulatory authorities are preparing to adopt standards for AI, like the EU law on AI currently being drafted. Yet it will take time before these are fully in force, and before AI best practice is established.

Start with small steps
Logistics companies should weigh up the advantages of AI against the risks it entails. They can start off their AI journey slowly while they gradually gain more experience. Even this will give them an advantage over the competition, which as we know never sleeps. Success will come to companies that experiment with AI, utilise it, and incorporate it into their processes to offer more efficient and reliable services, as AI-based logistics orchestration will give them a competitive edge in the industry.

This article was first published in Handelszeitung.

Thank you to Ruben Principe, a key contributor to this article.