Design is an action that is characterized by enlarging the search space for possible solutions to an existing problem. However, the concrete form of the solution is not further defined: it includes artefacts as well as actions and ideas, but can also negate them by analyzing the problem. This paper analyses two different perspectives on a designing AI: On the one hand, AI describes agents who perceive their surroundings and try to solve their problems as best as possible. On the other hand, AI describes the ability of computer programs to imitate human approaches using skills such as learning. According to these two perspectives, problems must be perceptible and systematically learnable so that a designing AI is superior to both humans and their algorithms. Nevertheless, a designing AI won’t be a designer. This might lead to a democratization of design, as more people will have access to the creation of design artefacts through the tools, but also leads to the fact that parts of the design process will be subject to change. The use of AI in design comes to a limit, when the execution of design processes is essential and existing paradigms are to be questioned and newly developed.