The article that you will review is "Learning Scene Functionality via Activity Prediction and Hallucination". Given all we have seen in this class so far, you should feel technically competent to review the paper.
Here are two classic write-ups related to writing SIGGRAPH reviews, both by pioneers in the field of computer graphics: "How to get your SIGGRAPH paper rejected?" by Jim Kayia in 1993, and "Writing (SIGGRAPH) Technical Reviews" by Greg Turk in 2008. Although the questions in the reviews you will write are different now (they are more general), these past wisdoms still very much apply today.
Your review should be no shorter than 2 pages and no longer than 2.5 pages, including the questions prompts (in MS Word, 1.5-spacing, following this template).
Computers Do Not Make Art, People Do, Communications of the ACM, May 2020, Vol. 63 No. 5, Pages 45-48.
This is an interesting column about AI, creativity, and art. It is referred to as a "viewpoint", not a technical research paper. The critique you write should provide a critical assessment of the article's main point(s), the arguments presented, and the evidences provided. In the end, you may or may not agree with the main message, but should put forward thoughts that would support your position, either for or against.
There are a lot of online resources to offer useful pointers on writing article critiques. Here is one guideline, and here is another link, where there is a sample critique. You are advised to follow the guidelines provided within.
The grading of your critique will put 80% weight on the quality of the critical assessments and 20% on the quality of the writing.