Artificial intelligence (AI) applications will have a lasting impact on the global economy in the years to come. Right at the front: The large digital platform operators from the USA and China. But even for medium-sized companies, there are fields of application in which AI solutions can be used sensibly today.
New success stories about applications of artificial intelligence (AI) can be found in the media every day. For example, AI systems find tumors on x-rays, compose pieces of music, compose news, optimize weather and flood forecasts, control autonomous cars, predict crimes and even win against professionals in the Asian strategy game Go. But the streaming portal Netflix, for example, also uses AI to display individualized film recommendations to its 167 million users and thus bind them more closely to the offer. The application potential of the technology seems inexhaustible – or as the founding editor of the American technology magazine Wired, Kevin Kelly, put it back in 2016: “Just as electrification catapulted all technologies to a new level 150 years ago, so will deep learning”.
This is shown not least in the enormous research and development expenditures of the big American player’s Google, Apple, Facebook, Amazon, and Microsoft (“GAFAM”) as well as the dominant Chinese groups Baidu, Alibaba, and Tencent (“BAT”) in this area. The same companies are also major AI venture capital investors and buyers of startup companies in the AI sector. The Americans and Chinese also dominate the number of patents in this area, with the number of AI patents skyrocketing since 2013. This is shown not least in the enormous research and development expenditures of the big American player’s Google, Apple, Facebook, Amazon, and Microsoft (“GAFAM”) as well as the dominant Chinese groups Baidu, Alibaba, and Tencent (“BAT”) in this area. The same companies are also major AI venture capital investors and buyers of startup companies in the AI sector. The Americans and Chinese also dominate the number of patents in this area, with the number of AI patents skyrocketing since 2013. This is shown not least in the enormous research and development expenditures of the big American player’s Google, Apple, Facebook, Amazon, and Microsoft (“GAFAM”) as well as the dominant Chinese groups Baidu, Alibaba, and Tencent (“BAT”) in this area. The same companies are also major AI venture capital investors and buyers of startup companies in the AI sector. The Americans and Chinese also dominate the number of patents in this area, with the number of AI patents skyrocketing since 2013.
What Is Machine Learning?
Artificial intelligence is not a self-contained technology. Rather, it is a field of research in which various methods from different disciplines are summarized under the term AI. The methods that are currently making headlines and celebrating success in more and more application areas stand for clearly delimited sub-areas of AI. In essence, it is mostly about the forecasting ability of the systems, and that is very different areas. To do this, they need large amounts of high-quality training data. The current AI applications rely almost exclusively on “machine learning” – that is, machine learning – and deep learning, a sub-discipline of machine learning. In machine learning, the AI system learns using examples and can generalize them after the learning phase has ended. To do this, algorithms build a statistical model based on training data. This means that the examples are not memorized, but patterns and regularities in the learning data are recognized. In this way, the system can also assess unknown data. In deep learning, so-called neural networks are also used as part of the “learning process”.
AI Frameworks Form The Basis Of Numerous Applications
In industry and research, relatively few “building blocks” dominate the development of artificial intelligence applications around the world. These frameworks contain large libraries of preconfigured algorithms. For example, the TensorFlow framework, which was published by Google in 2015, is widespread. The PyTorch program library from Facebook is also very popular. Both rely on the Python programming language. The ML.NET framework is still relatively young – an extensive software library for machine learning from Microsoft. Technically, it is based on the Microsoft technologies .NET and .NET Core. These and a few other frameworks can all be used freely under different variants of open source licenses.
Nevertheless, small and medium-sized companies have high hurdles to use machine learning in practice. The definition and development of a business case or an application scenario, the choice of the right framework and the right algorithms, the testing of systems, and the final rollout in practice are not exactly routine tasks in the day-to-day business of a retailer. Last but not least, the right expertise from computer scientists or “data scientists” for such a project is difficult to get.
The Intelligent ERP System
The concept of the intelligent ERP system, also called ERP, has been discussed in the trade press for some time. Valuable and meaningful data can already be found in companies’ ERP systems. However, dealing with “abnormalities” has so far been rather reactive. In other words, a company could only react when a problem had already arisen. The added value of an intelligent ERP system consists, among other things, in the fact that it sends reports “by itself”, i.e. proactively, when it detects conspicuous developments. In the ideal case, for example, an impending loss of customers can be prevented before it occurs. In its current position paper “Artificial Intelligence and ERP”, the IT industry association Bitkom takes the view that that AI will become a key competitive factor both on the part of the ERP provider and on the part of the user companies The position paper also describes the path to an AI-supported organization, in which algorithms help with decision-making or, in the future, may even make decisions automatically.