April 2024

As already mentioned in an earlier blog post, we plan to gradually integrate AI applications into our research. First, however, we need to find out which tools match our needs and which AI applications are genuinely adding value to our services. That's why we launched a comprehensive AI test a few weeks ago. We are combining two testing modes: a broad-based one plus a targeted and structured approach. In this way, we hope to obtain as differentiated and complete a picture as possible. For example, we repeat searches that have already been completed and compare our previous results without the use of AI with the new results. The applications tested include not only classic AI chatbots (generative AI) but also tools tailored to our needs and AI that is integrated into databases that we already use. The main question is which AIs can support us in our research and analyses, for example by providing us with new sources, information and search terms.

Even though the test has not yet been completed, clearly some of the applications can support us in our daily work – whether they bundle news thematically, provide new inspiration or offer a quick overview of many or long documents. However, the limits of AIs are also becoming apparent, for example when texts are misinterpreted, summarised incorrectly, or context is presented inaccurately. 

This illustrates clearly, that in the end, critical questioning and strict guidelines for the use of AI remain important. And, in any case, we will continue to rely on our own words.