Nils Urbach, Professor for Business Informatics and Strategic IT Management at the University of Bayreuth, Peter Hofmann, University of Bayreuth and Dominik Protschky, University of Bayreuth bet that in five years' time, anchoring in the company will be one of the biggest challenges in the successful application of artificial intelligence.
A 35-year lap record at the Nürburgring is undercut by an impressive 51.58 seconds thanks to the use of Artificial Intelligence (AI) laps. The machine now beats humans at poker and even masters bluffing. Hardly a week goes by without a new breakthrough in AI racing being announced.
It is precisely these technical breakthroughs that are once again generating high expectations of the potential of the technology. Those who do not want to be left behind in the AI race face the challenge of anchoring the technology and its applications in the context of the entire company. CxOs will soon wish they had thought about their AI strategy earlier.
However, AI is not a new topic. The idea of developing a self-learning machine has aroused the interest of scientists and practitioners since the 1940s. AI serves as a generic term to summarize a variety of methods and applications that aim to perform tasks that typically require human intelligence. For example, the Turing test formulated by Alan Turing in 1950 is still the basis for the discussion of machine intelligence today.
Milestones in the history of AI
Since then, the quest to create intelligent machines has manifested itself in several milestones, including in the recent past the victory in chess against the world champion Kasparov by IBM Deep Blue (1996), the win of the quiz show Jeopardy by IBM Watson (2011) or the victory against the current world champion in the game Go by Deepmind Alpha Go (2017). While AI research has gone through many heights, disappointments due to exaggerated expectations also led several times to phases of less interest and investment, which we also call AI winter.
At the moment, expectations seem to be running high again, so that the question is obvious: Is the next AI winter already around the corner? In contrast to the developments of the last century, AI has already found its way into our private and business lives in many different ways. For example, the translation service DeepL, which is based on Deep Learning, already enables amazingly error-free translations. And the language assistant Google Duplex can independently call a restaurant to -reserve a table.
Creating knowledge from data
Machine Learning, in particular, an important approach to AI development, is used in many applications to learn knowledge or skills from data. Despite the prevailing euphoria, all these applications can be assigned to the weak AI thesis.
In contrast to the strong AI thesis, which aims at an almost complete imitation of human thinking and acting with machine systems and thus at a generic intelligence, the weak AI thesis follows only concrete application problems to be solved intelligently. Google Duplex, for example, knows how to make a phone call in a deceptively human manner, but it is better not (yet) to entrust the service with driving your own car.
This article was published on cio.de