The elephant in the lecture hall

Let's address this directly. Yes, students are using AI. Yes, universities know. Yes, there's a massive grey area between "helpful study tool" and "academic dishonesty." And no, your university's AI policy probably doesn't clarify things as much as it should.

Here's the line, stated as simply as I can manage: if using AI helps you understand the material better, that's studying. If using AI helps you avoid understanding the material, that's cheating. Everything else is detail.

A calculator helps you do maths faster, but if you use it without understanding what the numbers mean, you've learned nothing. AI is the same. It's a powerful tool that amplifies whatever approach you bring to it. Bring genuine curiosity and it accelerates your learning. Bring laziness and it accelerates your failure (even if it takes a term or two to catch up with you).

The tools worth knowing about

chatgpt and claude are general-purpose AI assistants. Both are excellent for explaining concepts, generating practice questions, and talking through ideas. Claude tends to give longer, more nuanced explanations. ChatGPT tends to be more concise and direct.

notebook-lm is Google's document-based AI research tool. Upload your lecture notes, textbook chapters, and readings, then ask questions about that specific material. This is arguably the most useful AI tool for students because it only answers from your sources, not from the general internet. No hallucinated facts, no information your course doesn't cover.

perplexity is an AI-powered research tool that searches the web and cites its sources. Useful for finding additional reading, checking facts, and understanding topics that your lecture notes don't cover in enough detail.

Study workflows that actually help you learn

Workflow 1: The concept explainer

You're reading a textbook chapter on macroeconomic policy and you hit a paragraph about quantitative easing that makes no sense. Normally you'd read it three more times, still not understand it, and move on hoping it won't be in the exam.

Instead: Copy the paragraph into Claude and say: "Explain this concept to me in simple terms. Then explain it in slightly more technical terms. Then give me a real-world example from the last 5 years."

Claude will give you three levels of explanation. Start with the simple one. When that clicks, move to the technical one. The real-world example anchors it in something concrete. This mimics how a good tutor teaches: scaffolding understanding from simple to complex.

Why this is studying, not cheating: You're working to understand the material. You're just getting a better explanation than the textbook provided. This is no different from watching a YouTube video about the same topic or asking a classmate to explain it.

Workflow 2: The practice question generator

The best way to test your understanding is practice questions. Textbooks usually have a few at the end of each chapter. You need more.

The prompt: "Based on this material [paste your notes], generate 15 exam-style questions. Include: 5 multiple choice, 5 short answer, and 5 essay-style questions. For the multiple choice, make the wrong answers plausible, not obviously incorrect. After I answer each question, tell me if I'm right and explain why."

Then actually answer the questions. Type out your answers. Let the AI mark them. Pay attention to what you get wrong. That's where your knowledge gaps are.

The advanced version: "Generate questions that test common misconceptions about this topic. What do students typically get wrong about [subject], and create questions that would expose those misunderstandings?"

Workflow 3: The debate partner

For humanities and social science subjects, understanding isn't just about knowing facts. It's about evaluating arguments, considering perspectives, and building your own position.

The prompt: "I'm writing an essay about [topic]. My current thesis is [your argument]. Play devil's advocate. Challenge my argument from the strongest possible opposing position. Be specific and use evidence."

This is gold for essay preparation. Instead of only researching evidence that supports your argument (confirmation bias), you're forced to contend with the strongest counterarguments. Your essay will be better because you've already addressed the obvious objections.

Follow up with: "Now help me strengthen my original argument to address the points you raised." This back-and-forth is essentially what a tutorial discussion should be, and it's available at 2am the night before your essay is due when your tutor decidedly is not.

Workflow 4: NotebookLM for document-based studying

This is the workflow I recommend most highly for exam revision.

Setup (10 minutes): Create a notebook in notebook-lm Upload all your lecture slides, notes, and required readings for one module Let it process everything

Daily revision (30 minutes): Ask: "What are the five most important concepts covered in these materials?" For each concept, ask: "Explain [concept] using only the information from my sources. Include specific page references." Ask: "What connections exist between [concept A] and [concept B] based on these materials?" Ask: "Generate five exam questions based on these sources, focusing on areas where different sources present different perspectives."

Before the exam: Ask: "Create a comprehensive revision guide from these sources. Organise by theme, not by lecture date." Generate an audio overview and listen to it during your commute or while doing chores. Having the material explained in podcast form is surprisingly effective for retention. Ask: "What topics appear across multiple sources? These are likely important themes the examiner will focus on."

Why NotebookLM specifically: Because it only uses your uploaded sources. When you ask ChatGPT a question about your course material, it might answer using information from the wider internet that your lecturer didn't cover or doesn't agree with. NotebookLM stays within the bounds of what you've been taught. For exam revision, that's exactly what you want.

The things you should never do

I'm going to be direct about this because the consequences are genuinely serious. Academic misconduct can result in failing a module, failing the entire year, or being expelled. It's not worth it.

Never submit AI-generated text as your own writing. Not "lightly edited" AI output. Not "AI wrote the first draft and I rewrote it." If the structure, arguments, and phrasing originated from an AI, it's not your work. Use AI to understand the material, brainstorm arguments, and check your reasoning. Write the actual essay yourself.

Never use AI during closed-book exams (obviously). But also: don't use AI to memorise answers to predicted questions. The point of an exam is to test whether you understand the material well enough to apply it to unfamiliar questions. Memorised AI-generated answers fall apart the moment the question is slightly different from what you prepared.

Never rely on AI for specific facts without verification. AI tools hallucinate. They make up citations, invent statistics, and confidently state things that are wrong. If you include a fact from an AI in your essay, verify it from a primary source. Your lecturer will check, and "ChatGPT told me" is not a defence.

Never use AI to write code for programming assignments without understanding it. Your lecturer will ask you to explain your code. If you can't walk through every line and explain why it's there, you didn't write it. More practically, if you can't code without AI, your exam results will reflect that, and you'll have a degree in a subject you can't actually do.

The ethical framework (simple version)

Before using AI for any academic task, ask yourself one question: "After using this tool, will I understand the material better or worse than before?"

If better: go ahead. You're studying. If the same: it's probably a waste of time but not harmful. If worse (because you skipped the learning): stop. You're cheating yourself, and possibly cheating academically.

Your degree is supposed to mean you know something. If you use AI to avoid learning that something, you're paying tens of thousands of pounds for a piece of paper that misrepresents your abilities. That catches up with you eventually, usually at the worst possible moment, like your first week in a job where people expect you to actually know what your degree says you know.

The honest conclusion about AI and education

AI tools are the most powerful study aids ever created. A personal tutor available 24/7 that can explain any concept at any level, generate unlimited practice questions, and help you think through complex arguments. Twenty years ago, only students who could afford private tutoring had access to this kind of support. Now it's free.

The students who use AI well will learn faster and deeper than any previous generation. The students who use AI to avoid learning will be worse off than if AI didn't exist, because they'll have the illusion of competence without the reality.

Which one you become is entirely up to you. The tools don't care.