I have a habit of reading course reviews before signing up for them. It helps me form an idea of its contents and decide whether to sign up and devote my time or not. I am seldom surprised much, but in the case of this course, I must admit I was amazed by the gap between the course rating and its content. I was aware, of course, of the 2-star review (by Irina G, posted on Nov 22, 2020) which, at that time, had by far the most votes: 18. The next most upvoted review (by Giora S, Jan 7, 2021) had 7 votes and it also gave only 3-stars. At that time, I was not alarmed by the fact that the most upvoted review giving 5 stars to the course had only 3 votes. Although it is somewhat surprising, I can understand that many people may feel good about a course just because it is easy to pass and, consequently, have a course like this have a rating of 4.9. But how does Coursera select top reviews with 5-stars and only very few upvotes is beyond me. This really makes me reconsider Coursera's rating system. I fully agree with Irina G's review. Could add few more observations of mine, but I do not see the point. I really appreciate the mission of Coursera, the effort made by the lecturers to prepare a course and the fact that we can listen to it for a comparatively very low price. But having Laurence read pre-written text from a screen with generic statements, explaining virtually nothing other than reading the code taken in large part from Keras tutorials which, at least, have explanations, is quite surprising. Even more so considering its rating is pumped up to 4.9.