Yes, AI did fail and will keep on failing, because its human who is teaching and training these models. Anywhere, if AI or Artificial Intelligence is failing, which means there is some mistake in coding, algorithms, or a defect in manufacturing. Because the AI model is not performing the way it's intended, means there is a defect in the back end. Because these machines or computers are good at following instructions and will never fail to make mistakes or laziness in performing them. Maybe the AI model has learned something different which the instructor did not mean to or was not clear enough to differentiate, thereby errors creep into the system.
There have been several reports, of AI not performing the intended tasks, somewhere crossing the lines, involving in dangerous activities it was not intended to, reading, or understanding differently than the way it was made to. But these are mistakes and errors which the tech companies are looking at closely and resolving them to make generalized AI a reality, one by one.
If you are serious about learning AI, check out the Artificial Intelligence online course from Intellipaat. Also, check out the AI tutorial for beginners to help them understand better.