Overall the class was pretty good. The lectures were entertaining and the homework assigned wasn't too bad (though some of the coding assignments were pretty difficult). My main complaint about this course is that the final was supposedly concept-based, but IMO had a lot of definition/memorization based questions.

Expectations for the class are clearly conveyed and the study guides for the midterm and final are very useful

Prof. Van den Broeck seems to have improved greatly from previous quarters. I did not see any issue with his lecture style -- his lectures were pretty interesting and I thought he explained the material clearly. The only thing I would say is that he went by a little too fast during live lectures for me to keep up, so I had to supplement with the recordings from F21, which remain an excellent resource for learning the material.

Past reviews say that the homeworks would take only an hour or so if we were allowed to use a modern programming language, but it seems he changed the homeworks significantly once the programming language for the course was switched to python. There were four homeworks in total (there were supposed to be five but we ran out of time in the quarter for the fifth one to be posted), and each took about 3-5 hours. All but one of the homeworks were fairly challenging -- rather than worrying about Lisp as in previous years, students can now grapple with actual artificial intelligence concepts.

The exams too seem to have been improved. As the previous reviews mention, the midterm is fairly straightforward, with around a 75 average. However, there were some pretty serious errors with the grading, but most of them were caught and fixed by the TAs (I did have to go to office hours to see my exam in person to catch all the mistakes though). Fortunately, the final exam, while tough, was nowhere near as difficult as the past reviews make it out to be. One review mentioned that if a student knows everything covered in the lectures from top-to-bottom, they could maybe get a 75% on the exam maximum. However, this year, that seemed to be more like 90%. There were only a few questions that were unreasonable given what was covered in lecture. Speaking of which, the professor strongly recommends the textbook for studying from the exams, but I personally found it more than sufficient to just study the lectures in-depth. That being said, the amount of content we were expected to know was frankly unreasonable considering the closed-notes nature of the exams. I ran some calculations and found that each CS 161 lecture covers 2.5x the material of a typical CS 181 lecture (which was closed-note) and almost as much material as a typical CS 143 lecture (which was open-note). The lectures were so dense that studying each lecture took me 4 hours (meaning that it was essentially a full-time job just to study two lectures per day). However, that mostly applied to the lectures in the first half of the course (before the midterm); I found the ones in the latter half to be far more reasonable in terms of the amount of content covered. Nonetheless, no need to stress if you can't study every single detail, as there is a curve applied at the end of the course. For reference, I received ~93% raw score which was curved up to an A+.

My main gripe with the course is that there were very few resources for students to get help. The professor did not hold regular office hours, which I consider to be the bare minimum. To speak with the professor, you need to schedule an appointment with him. The TAs were also apathetic and unresponsive. Aside from that, however, I found the course to be enjoyable overall, as the workload was significantly lighter compared to other CS classes and the material was absolutely fascinating. Studying the material in this course even helped me significantly in an interview, causing me to get a 190k TC offer! Given the improvements made to this course, I would definitely recommend it.