We have some interesting discussions elsewhere about KI Apps nor telling the truth. Reimann did the most interessting experiment, exposing three KI applications to a single set of questions. Taking the human expert answer to this question revealed, that two out of three KI Apps did not „know“ or „find“ the answer. This rises fundamental questions about the „truth“ of KI Apps answers.
The case in German language:
https://www.schadenseminar.de/2026/01/18/pr%C3%BCfung-bestanden-trotz-ki/