Basic knowledge Artificial Intelligence in logistics: examples, opportunities, risks

In this article you will learn what Artificial Intelligence is, how it works and what opportunities it offers for logistics and supply chain management!

AI provides for increasing process efficiencies, supplies most accurate prediction models and enables an unprecedented ability to adapt to changing markets - no surprise that last year more than $ 40 billion were invested in the research of Artificial Intelligence globally. Many people regard it as one of the most important growth drivers for logistics over the next few years and as the most important key to competitiveness.

What is Artificial Intelligence (AI)?

Definition of Artificial Intelligence: The term Artificial Intelligence refers to the ability of machines to interpret different problems and to independently develop suitable solutions. Instead of working through rigid algorithms AI-Machines make their decisions afterwards and can therefore acquire a well-founded wealth of experience. With its help, they can develop ever better solutions and even make predictions.

A somewhat broader explanation for Artificial Intelligence is given on the website Understanding Databases: "Artificial intelligence is a computer science subdivision. But it goes much further and has been shaped by psychology, neuroscience, philosophy, communication sciences, mathematics and linguistics".

While experts can determine human intelligence relatively easily, the question as to when a machine is considered intelligent is much more difficult to answer. It has been a matter of hot debates in research for many decades. An assessment instrument that is recognized by scientists worldwide is the Turing Test. To date, however, no machine has been able to pass it - and that will probably remain so for some time to come.

Science, by the way, prefers to speak of "Deep Learning" or "Machine Learning".

What is the basis of artificial intelligence?

Three things are needed to use and develop artificial intelligence: high computing power, large amounts of data and intelligent algorithms. Since the availability of these resources is drastically increasing today, Artificial Intelligence is currently experiencing a real boom.

• Computing power increases drastically: As recently as 1995, the world's most powerful supercomputers were able to perform just 100 billion computing operations per second - today any good mobile phone can do so. Even the energy consumption vof computers and the corresponding material costs are much lower than at the time. For example, a modern smartphone needs only a 10,000th to a 100,000th of the electrical power of earlier supercomputers.
  
  And an end to this development is not in sight: According to the renowned science journalist Ulrich Eberl, it is likely that the performance of microchips will increase by a factor of 1000 by 2040.

• Big Data is gaining momentum: Increasing networking along the supply chain is making sensors smaller, better and cheaper - from RFID to vibration and heat sensors. In addition to the veritable data explosion on the Worldwide Web, this ensures the generation of incredible amounts of data (keyword Big Data). Data availability is also constantly increasing: While in the past chaotic databases had to be managed, today sorting by algorithms and technologies such as blockchain is used in many cases.

• Better Algorithms: The pace of software developments is also steadily increasing. The programming process for ever faster and better algorithms that AI applications can ultimately use is progressing steadily.

How does Artificial Intelligence work?

Similar to humans, self-learning machines do not possess their abilities immediately after "completion“ but have to learn (and optimize) them step by step. In doing so, they proceed - as we do - according to the trial and error method.

The first question that was to be answered by AI algorithms was: „"How can software independently learn to recognize a cat in a photo?" A look at the solution to this problem illustrates how systems based on artificial intelligence are basically constructed and how they work:

• 1. Step: The software used for solving the problem consists of several modules. Each individual module examines its own variable (such as the mouth, ears or eyes) in the image.

• 2. Step: During the learning process, the user adds several graphics to the software. Included are pictures of cats as well as other (non-cat) photos.

• 3. Step: Each software module decides individually whether the variable scanned in the image is to be assigned to a cat or not and accordingly gives the feedback "Yes, it is a cat image" or "No, it is not a cat image". For the first photo, the system makes a purely random decision.

• 4. Step: After the decision has been made, the program uses the metadata of the image to determine whether it has actually been categorized as a cat image by the user. The modules that have made a false statement will be adjusted to allow them to correctly identify the photo in the future.

• 5. Step: A further image is uploaded to the Artificial Intelligence. The system modules are now a little smarter and make their statements based on the experiences they made with Picture 1.

• 6. Step: Once the application has examined the second metadata, it readjusts all modules that were wrong this time. All modules can now correctly judge both the first and the second photo.

• 7. Step: The software is gradually refined and enhanced with the help of many more images so that it can make ever more reliable statements.

Business Software

Enterprise Resource Planning (ERP) — definition, types and benefits

The quality of Artificial Intelligence depends very much on the extent and quality of the training data. Only if this information is correct and complete can the software make the right decisions and weight the statements of the individual modules in a meaningful way.

What happens when artificial intelligence is provided with a qualitatively poor database for learning purposes, Microsoft had to painfully experience with its Chatbot Tay. Just a few hours after its launch, it started insulting Microsoft followers on Twitter, denying the Holocaust and pleading for the construction of the wall between the USA and Mexico. The users had "fed" it with dubious information and thus provided the Artificial Intelligence with the basis for its inflammatory statements.

Compared to humans, machines (as expected) learn much faster: After just three weeks, an AI application from Google subsidiary Deepmind mastered the millennia-old game of Go better than any human has ever done before.

Where is Artificial Intelligence used?

Decision-makers are considering artificial intelligence as one of the key technologies of the future. According to a potential analysis by Hermes, 20 % of all companies in the logistics and transport industry already use these kinds of applications. After all, another third is planning to use AI in the next few years.

• Artificial intelligence is most frequently used in the field of robot-controlled process automation. For example, the input of data for ERP systems can be considerably simplified by Artificial Intelligence or software robots.

• AI is also in demand when it comes to intelligent sensor technology, such as real-time data acquisition.

• However, the most controversially discussed field of application for Artificial Intelligence in logistics is in self-driving vehicles. With the help of self-learning algorithms, they can perceive their environment, react to traffic in real time and plan efficient routes. The software of the vehicle analyzes the planned transport route, compares it with the actual traffic volume and traffic jam messages and then calculates the fastest route.
  
  This considerably reduces operating costs and congestion. Autonomous systems are also able to operate 24 hours a day, making them an attractive way to counter the lack of drivers. Artificial intelligence ensures that empty runs are avoided. This means fewer trucks on the roads, a reduced risk of accidents, fewer traffic jams and more transparent prices.

• In the context of Predictive Maintenance, Artificial Intelligence can predict machine failures and instruct employees how to fix the problem. In an automatic small parts warehouse, for example, the software recognizes certain parameters during operation of the storage and retrieval machine (such as a grinding noise) that are typical for machines that are about to fail. It then provides early notification to the user to prevent downtime, reduce costs and increase the productivity of logistics systems.

• Another promising application of Artificial Intelligence is in warehouse technology. AI algorithms specify which goods are to be stored in which logistics warehouses. The software takes into account the consumer behavior for different goods at different locations. For example, it may find that product A is more in demand in rural areas and has to be stored in logistics warehouses there and product B in distribution centers close to cities.

Further areas of application for Artificial Intelligence are currently in the areas of crime prevention, medicine, health and voice commands.

Logistics properties

What are logistics properties? Definitions, market & requirements!

Risks of AI: Will artificial intelligence soon replace humans?

One of the biggest fears among professionals and executives about artificial intelligence is that it could sooner or later make people completely superfluous. However, experts such as Evi Hartmann, Professor of Business Administration (especially Supply Chain Management) at the University of Erlangen-Nuremberg, agree: “On the one hand, jobs will be eliminated by this technology, but on the other hand it will also create new business models, services, products, industries and job descriptions. If, for example, with the help of artificial intelligence trucks move autonomously at some point, the former truck driver can conduct more demanding logistics tasks." Artificial Intelligence will probably revolutionize people's work, but not make them superfluous.

According to the Logistic Trend Index of the Munich Trade Fair Centre however, 70% of professional and managerial staff currently experience a predominantly negative attitude towards Artificial Intelligence on the part of their employees. To ensure that the interaction between man and machine can take place smoothly in the future, German companies still need to provide a great deal of training and education.

This article was first published byMM Logistik.

For more news visit our facebook page or twitter.

(ID:46212065)