Critical paper reading and writing are essential skills for researchers in their daily life. They require researchers to think critically about the paper content they read or write, usually by asking questions like “what is the motivation” and “is it convincing enough”. This project seeks to develop intelligent agents that can ask questions about the paper and give feedback to users to facilitate their critical thinking such as in paper reading and writing. The main challenges are that 1) the agents need to understand the task content to ask the right questions and give useful feedback and 2) they need to detect users’ focus (on task materials) to interact at the right timing. Machine learning approaches can be used to address these challenges from several aspects, e.g., question generation, reading comprehension, intent classification, text matching, etc.