What do we need intelligence for?

Why do we need artificial intelligence?


"AI systems that offer solutions for business analytics or intelligent automation have the potential to change almost any industry," says Hendrik Reese, Director at PwC, with certainty. "Increasing amounts of data combined with limited, classic IT systems and limited human capabilities call for new approaches to data analysis." AI achievements improve our lives in many ways, as Reese emphasizes: "Artificial intelligence can support doctors in diagnosing cancer early on and show optimized healing options. In addition to e-mobility, the automotive sector will also be massively influenced by AI-supported, autonomous driving. "The fashion sector will also benefit from intelligent technologies, as Stacia Carr, Director of Engineering at Zalando, emphasizes:" At Zalando, AI means that the focus is on people. We want to make sure our customers have the best shopping experience. For our employees, this means that they have the opportunity to use new technologies and thus develop themselves further. "


Stacia Carr works as Director of Engineering at Zalando, where he heads the “Sizing Team”, “a cross-functional team of software engineers, data scientists, business developers and fitting models that deals with product sizes and the right fit on the customer's body.” Help From artificial intelligence, mail order company Zalando is working towards ensuring that products in the store are labeled if their size deviates from the norm. “Customers get the right fit when they place their first order,” adds Carr. In the fashion industry, says the engineer, the surface is still being scratched: “We have spent a lot of time with 3D technologies and how these can help to improve our offer for customers.” The goal is clear: “Artificial intelligence should used throughout our company to improve customer satisfaction, scale products and work more efficiently. "


"Everyone is talking about AI expert systems such as chat or voicebots and are increasingly being used in customer service, for example," says Jan Morgenthal, Chief Product Owner Artificial Intelligence at Deutsche Telekom. According to the expert, this was due in particular to developments in the area of ​​deep learning from recent years. »The aim of these is to simulate human characteristics, such as the semantic understanding of written and spoken language. But ethics within AI is also an important trend: Large amounts of data are required for training AI technologies, so that sources of error can creep into the systems very quickly. ”This should be avoided. Morgenthal assumes that the “Robotic Process Automation” trend in its current form will decrease significantly, “because it is a bridging technology and so far has hardly anything to do with AI. However, it is already becoming apparent that there will be successor technologies that can be classified in the area of ​​›Cognitive Process Automation‹. "


Even if reinforcement learning, in which an AI learns in a highly simplified way according to the trial-and-error principle, according to Dr. Thomas Franz can still be used to a very limited extent today, believes the head of adesso's Technology Advisory Board: “We achieved unbelievable successes, for example in learning to play games. The poker AI ›Pluribus‹, for example, was able to defeat several human players at the same time. This shows that even highly complex games can now be learned by machines, so that humans are de facto inferior. "If it were possible to transfer reinforcement approaches more to work and everyday situations, that would be a significant step towards being flexible usable AI solutions, says Dr. Franz. Regardless of the AI ​​process, one can expect that the understanding of speech and images in everyday life will change and simplify our interaction with technical devices or the handling of annoying tasks: »Basically, AI will support decisions in more and more areas and make them more efficient Contribute to processes. I think we will see more and more individual AI, that is, artificial intelligences that provide specific solutions for specific technical, context-specific problems. "


"An important factor that slows the productive use of AI in the corporate environment is a lack of data," knows
Hendrik Reese, Director at PwC. “However, organizations are realizing more and more that they already have the necessary data - they just have to localize it and label it correctly in order to use it to add value.” Another brake on innovation is a lack of trust in AI. In order to develop the full potential of AI, international standards are needed that enable the certification of AI systems. "Using independent assessment procedures, we can then establish a common understanding and a suitable level of security in AI in both B2B and B2C interactions," concludes Reese. In conclusion, he adds: »I see an increasing dovetailing of AI with other technologies such as the Internet of Things, blockchain or quantum computing. A combination of technologies can be used to effectively implement use cases. In this way, new solution concepts can be created. "