AI and academic writing, paper by paper

The short version

AI writing tools are now part of how many students work, and most universities have stopped pretending otherwise. What they have not done is agree on one rule. One module lets you use a chatbot to plan and brainstorm, the next bans it outright, and a third asks you to declare every prompt in an appendix. So the useful question is rarely "can I use AI?". It is "what is this paper asking me to show, and does the tool help me show it or do it for me?" The pages below answer that for each kind of paper. A literature review has to show you know the debate, so a tool that invents citations is a liability. A thesis statement has to commit to a claim, so a tool trained to sound balanced pulls you the wrong way. A case study analysis has to show judgement, so the summary is the safe part to hand over and the recommendation is not. Each page sets out what the tools do well on that paper, the failure you are most likely to meet, and the line most course rules draw. None of it replaces your own course's policy. If your module handbook says something different, it wins.