Off the top of my head, some of the key ideas include:
* Affordances, that objects should have (often visual) cues that give hints as to how to use things * Mental models, that every design has three different models, namely system implementation, design model, and user model, and that the design model and user model should try to match each other * Gulf of Evaluation (the gap between the current system state and people's understanding of it) and Gulf of Execution (the gap between what people want the system to do and how to use the system to do it) * Kinds of Errors and how to design to prevent and recover from them, e.g. slips (chose the right action but accidentally did the wrong thing, e.g. fat finger) vs mistakes (chose the wrong action to do)
What's particularly useful about Norman's book is that these key ideas apply for all kinds of user interfaces, from command-line to GUI to voice-only to AR/VR to AI chatbot. I'd encourage you to think about this book in this kind of framing, that it gives you general frameworks for reasoning and talking about UX problems rather than specific practical solutions.