There was a time when the calculator was an innovation. In this case, however, it quickly became clear that users could trust the device and nothing could go wrong. Today, it's all about artificial intelligence (AI), which can do much more than a calculator, but can also produce way more errors. Susanne Ambros from the IT company QualityMinds in Munich uses this example to vividly explain the problem: „You can no longer believe everything AI says.“ In contrast to a pocket calculator, AI has hallucinations.

But the problem is also that „we haven't practiced critical thinking towards computers at all," she adds. "Because we always thought they were infallible.“ But, as we all know, AI is only as good as the data it is fed with. That's why one of the biggest challenges associated with AI is using it correctly and checking its quality.

QualityMinds was one of many companies taking part in the IT Summit in Frankfurt, held on 26 June at the KfW premises. Around 400 industry experts and managers discussed current IT topics and challenges. Ambros addresses something that also concerns other participants: even if everything revolves around AI, the challenging issues in this context are more of a human nature.

It's no longer about introducing a new technology, according to industry experts. Generative AI (GenAI) and other innovations can seem irrelevant to some employees, but they need to be brought on board.

Udo Müller, Director Business Development Manager at IT consultant Zühlke, asked the panel about the experiences of the participating companies and whether employees are „tired of technology“. The answers show that the motivation of many employees when using new technology is limited, and that a lack of training and involvement leads to additional frustration.

This is another reason why the experts believe it is particularly important to use AI specifically to solve particular problems. The standard is not enough, says Müller. „Introducing ChatGPT is not enough. You need a specific problem to start with.“ What comes out of the box is able to cover around 70% of human knowledge. But it's about getting as close to 100% as possible.

Software developer MaibornWolff from Munich is familiar with the problem from a company perspective: customers still often come with very unspecific requests. „Along the lines of: everyone is doing AI, now we want it too,“ summarises Carina Wagner, who works in Business Development. However, there is a discernible trend that requests and ideas are gradually becoming more specific. MaibornWolff develops and tests IT solutions. For customers with AI inquiries, the company first carries out a check to see whether they are even ready for AI. It looks at how the data is processed and where it is stored. If there is still chaos here, you have to start there first. Once again, the focus is on the data.

At the IT summit, the topic of hyperscalers and independence (digital sovereignty) naturally also came up. After all, Europe is not only dependent on the cloud, but also on artificial intelligence. Alexander Schroff from Publicis Sapient, a consultancy firm for digital business transformation, clarified that "we are not going to build new hyperscalers in Germany.“ Rather, the aim is to integrate the existing offerings into the established regulatory corset.