My thoughts exactly. This is just a case of an ML algorithm describing the face it sees on the Martian surface[0], kind of like a "machine vision Turing test." It's not that it isn't interesting; it's just that it's kind of the expected result when you train a neural net on zillions of images, some of which include faces.
---
[0]: https://www.space.com/11947-photos-mars-illusions-martian-fa...