There are two perspectives to answer this question; producers and consumers.

From the producers i.e., researchers' perspective, most of the existing AI models are based on statistical inferences of which most of the researchers have a thorough understanding. The same foundation proliferates into deep learning. Although the learning process of a deep learning algorithm is still a black box due to the enormity of the parameters (which usually are in millions), we as scientists are still able to comprehend the holistic training process. The detailed learning mechanism is of interest to the "Explainable AI" research community. Hence, in this perspective, our ability to comprehend artificial intelligence will not diminish over time.

From the consumers' perspective i.e., the general audience, the awareness of the AI techniques will still increase as the implications of AI embed more in daily life. Although the general consumers will not have a detailed understanding of the internal workings of AI, they will have an enhanced holistic awareness of how it works in general. Hence, even their ability to comprehend artificial intelligence will rather increase over time.

With the emergence of Explainable AI (XAI), the black-box nature of the AI model has been somewhat explainable. However, as the models have an exceptionally high number of parameters, it becomes virtually impossible to explain them.