Artificial intelligence projects often fail – and have earned the technology a reputation for being compelling in theory but difficult to apply in practice.

Wrongly so, believes Holger Mai from the consulting firm Accenture, who sees the reason for failed projects mostly in the wrong approach by managers.

"Of course there are also technical difficulties, but that's the minor part.

The problem usually lies in the management,” reports Mai in the FAZ podcast “Artificial Intelligence”.

The problem begins with the selection of the data: "In one case, the management had made the requirement to include 5000 data sources.

That was impossible.

80 percent of the value could be achieved with just 100 data sources,” says Mai.

“We make AI too complicated.

The pattern can be observed again and again".

Another mistake is the lack of scalability: Even if the pilot project worked, many companies stumble when the technology is to be rolled out in the company.

Management often makes the mistake of not thinking about this scaling from the start.

Another popular mistake: The “owners” of the data silos in companies are often not willing to make their data available for AI applications.

Here, too, management is required to define clear guidelines so that the best available data can be used.

However, anyone who has overcome the beginner's mistakes can noticeably increase the competitiveness of their company with AI.

For example, recognizing consumer trends or crises earlier or automatically thwarting attempts at fraud.

The episode is part of our podcast "Artificial Intelligence".

He explores the questions of what AI can do, where it is used, what it has already changed and what contribution it can make in the future.

With Peter Buxmann and Holger Schmidt, the FAZ brought two proven AI experts on board for the podcast: Both research and teach the potential of AI and its effects on the economy and work at the Technical University of Darmstadt.

Peter Buxmann holds the chair for business informatics and has been dealing with the applications of AI, digital transformation and data-based business models for many years.

His podcast partner Holger Schmidt is a digital economist, speaker and author.

His core topics are AI, platform economy and digital business models.

In each episode, the two hosts take up a new aspect of artificial intelligence, explain connections and provide precise classifications.

The episodes are around thirty minutes long and appear monthly on the first Monday.