Smash Your Uni Revision: Master AI Prompts & GPT Hacks

AI has become a hot topic among students, and I often get asked how to integrate it effectively into study routines. This guide aims to demystify some of the best practices I’ve picked up, so you can optimise your own revision process. Whether it's automating repetitive tasks, generating summaries, or create revision materials, AI has become an essential part of my uni life. Let’s explore how.

Framework & Prompting Tips

Getting the right output from AI largely comes down to how you frame your prompt. In fact, I rarely write a complete prompt by myself - every sample prompt you’ll see later in this blog was written by GPT, with a lot of direction from me. Think of it as a collaborative process: you provide the vision, and the AI fleshes out the details.

RISEN Framework

Don’t get me wrong - I hardly ever stick rigidly to any single framework when creating prompts, but simply being aware of optimal prompting can significantly enhance the quality of your AI outputs. For complex, multi-layered tasks, frameworks like RISEN offer a clear, step-by-step approach that ensures depth and precision. While variations of RISEN exist, the core principles remain the same:

1. Role: Define the AI’s expertise (e.g., “You are a marketing strategist”).

2. Instructions: Specify the task (e.g., “Create a campaign for eco-friendly products”).

3. Steps: Break the task into actionable phases (e.g., “First analyse trends, then draft copy”).

4. End Goal/Expectations: Clarify the desired outcome (e.g., “Increase engagement by 20%”).

5. Narrowing/Novelty: Set constraints or encourage creativity (e.g., “Exclude TikTok” or “Propose unconventional ideas”).

Greg Brockman’s o1 Prompt Framework

I have always follwed RISEN, but with new models and model design, knowing different frameworks is important. Developed to minimise hallucinations and improve accuracy, this four-pillar structure helps you produce more precise queries for reasoning models:

1. Goal: State an explicit objective (e.g., “Compare Arabica vs. Robusta coffee”).

2. Return Format: Specify the output structure (e.g., “Bullet points with taste profiles”).

3. Warnings: List any constraints (e.g., “Exclude rare coffee varieties”).

4. Context Dump: Provide background details (e.g., “Beginner-friendly, low acidity”).

Comparing GPT Models: When to Use Which

Understanding the nuances among various GPT models can significantly enhance your university study and revision strategies. Here's a straightforward breakdown of each model’s strengths and best applications for academic tasks:

GPT-4o

• Ideal for coursework or research projects involving multiple types of content—like text, images, graphs, or translations. With the new reasoning models, I use GPT-4o alot less for university (it still plays a big role in our AI systems). Now, I really only use it for specialised assistants (or quick questions), which I'll cover later on.

O3-mini

• Best Use Cases:
  Great for technical assignments, coding tasks, or STEM-related queries. It’s especially effective when you need quick answers or solutions, such as debugging code for computer science projects or rapidly processing large datasets for lab reports. I generally use it for coding, or when I have ran out of o1 messages (50 a week with ChatGPT Plus).

• Distinctive Features:
  Provides efficient, precise reasoning for technical questions, making it ideal when speed and accuracy are priorities in coursework or revision.

O1

• Best Use Cases:
  Exceptional for structured, concise outputs where logical precision and rapid response matter most. Ideal for tasks such as quickly summarising dense theoretical concepts, breaking down complex academic arguments, or producing clear, structured notes for revision. I have two specific/practical use cases below

• Distinctive Features:
  The o1 model excels due to its structured, step-by-step reasoning process, mirroring a systematic, human-like thought approach. This capability significantly reduces the likelihood of errors or irrelevant content, making it especially useful for complex yet concise tasks like evaluating economic theories, drafting policy recommendations, or analysing historical events in a structured manner. It’s also well-suited for constructing effective flashcards, outlining essays quickly, and summarising intricate lecture materials.

GPT-4.5

• Best Use Cases:
  A newer model still under development, GPT-4.5 aims to combine detailed reasoning with rapid response times. While I haven't extensively tested it myself, it’s worth watching as it could soon become very useful for academic tasks requiring both thorough analysis and efficient delivery.

• Distinctive Features:
  Promises improved clarity in responses, better understanding of nuanced academic prompts, and potentially fewer inaccuracies compared to earlier models.

Student-Focused Tips & Tricks for Study & Revision

1. Lecture Notes Transformation
   It’s not always possible to capture every crucial detail in a lecture. Sometimes you don’t have time to re-watch the entire recording, or you need a quick refresher on past modules. That’s where AI comes in. By feeding your lecture audio transcripts or notes into a GPT-powered prompt, you can generate comprehensive outlines and bullet points. It’s a major time-saver if you’re working through older lectures or just want to ensure you haven’t missed anything. Below is the prompt I use (with o1), it has changed alot over time, I probably change it slightly every time I use it - trial and error is so important in finding the output you want/need.

   "You are an expert at transforming lecture content into clear, structured, and comprehensive notes while ensuring all academic content is retained. In this task, you are provided with two key pieces of information: an auto-generated lecture transcript and the lecture slides (attached).

   Your task is as follows:

   1. Generate Detailed Notes from the Transcript:

      • Content Retention: Retain every academic concept, explanation, case study, and argument from the transcript. Do not exclude any academic content.

      • Gibberish Awareness: Note that the transcript may include some gibberish at the start, end, and at one point in the middle due to breaks in the lecture. While these parts should be handled carefully, ensure that all legitimate academic content is preserved.

      • Structured Format: Organise the notes according to the order of the lecture topics. Use clear headings for each topic or section. Under each heading, use bullet points to list key points, definitions, examples, and explanations.

      • Depth and Clarity: Summarise effectively by condensing the transcript while preserving the depth of content. Provide brief, complete explanations of each concept.

      • Error Handling: If any part of the transcript is illegible or contains obvious errors due to the automated generation process, infer the most likely intended meaning based strictly on the context of the lecture, without omitting any examinable academic content.

   2. Cross-Reference with the Slides:

      • Verification: Carefully review the lecture slides and cross-reference them with your notes. Although the transcript will always have more content, it may include some mistakes.

      • Corrections and Additions: Identify and correct any errors or misinterpretations in the transcript content by using the slides as the authoritative source (e.g. Names and dates are occasionally wrong in the transcript). For any academic content present in the slides but missing or misrepresented in the notes, integrate the correct information into the corresponding section.

      • Seamless Integration: Ensure the final notes are fully self-sufficient, so that any student could use them as their sole study resource for a 100% exam score.

   Output Requirements:

   • The final notes should be fully self-contained and organised in the order of the lecture.

   • Do not introduce any external information—only process and structure what is provided in the transcript and slides.

   • The final document must be written in clear UK English, with appropriate UK spelling and punctuation.

   • Use headings and bullet points to enhance readability and organisation.

   • Your output must be comprehnsive, capturing all acadmeic content. If a student were to take an exam, they should be able to obtain 100% using your notes alone.

   By following this process, your output will first capture every detail from the provided transcript, handling any gibberish appropriately, and then be refined using the slides to ensure complete accuracy and coverage of all examinable content.

   Context: The lecture is from the Economic Development module within an Economic Geography course."

2. Course-Specific GPTs
   This is great tool for writing essays and doing revision. With GPT Plus, you can create your own custom GPTs. You can upload all your lecture materials, slides, and readings into a custom GPT and transform how you access course content. Instead of rummaging through multiple PDF files or notes, you can simply ask your course-specific GPT a direct question - like clarifying a concept or relating one lecture’s content to another - and get immediate, context-rich answers. This approach is especially useful if you’re juggling multiple modules or fields of study at once.

   

   I also use Notion for a similar reason, Notions AI is great for finding specific concepts or theories buried within your notes and readings. This makes essay writing and revision more streamlined, especially when you need to quickly cross-reference ideas across multiple lectures or find niche details without sifting through pages of content.

   

3. Flashcard Generation
   Of course, making your own flashcards and revision resources is part of the revision process, so keep that in mind when using AI to speed up the process. With that little disclaimer out the way, I can guarantee you I'm not writing one flashcard this year, AI writes them better than I would in 1% of the time - an entire module worth of flashcards in 10 minutes (9 minutes of which is downloading lectures and writing a good prompt), something that usually takes days. All you need to do is provide your study notes and readings, and GPT can create question-and-answer pairs. Then, you can edit the flashcards so they feel more personal and aligned with your revision goals. I specfically use Quizlet, so you'll notice I have asked it to structure in an x = y format, this means I can import it directly into quizlet as a set rather than having to do one at a time. I also only wrote this yesterday, to test its effectiveness... the prompt will definitely change by the time I use it properly.

   "I have 10 documents of lecture notes covering regional economics, urban growth and decline, disruptive innovation, and related topics. Your task is to create a comprehensive set of Quizlet-style flashcards in the format “X = Y,” ensuring that a student who studies only these flashcards can achieve 100% in an exam on this material.

   Instructions and Requirements:

   1. Coverage: Include all relevant academic content from the 10 lectures. Ensure every major concept, definition, theory, model, key term, and important data point is captured.

   2. Format:

      • Use the structure “Term/Question/Prompt = Definition/Explanation/Quote.”

      • Vary the style when appropriate (e.g. fill-the-gap prompts, quotes from authors with date, or key models with an example).

   3. Organisation: You may group the flashcards by lecture or by thematic topics (e.g. Spatial Equilibrium, Agglomeration, New Economic Geography).

   4. Completeness: Assume the student has no other study material; provide enough detail so they can fully understand and recall each concept.

   5. Clarity and Conciseness: Each flashcard must be succinct but informative. If needed, break longer explanations into multiple flashcards for easier study.

   6. Academic Rigor: Where relevant, include names, dates, theoretical frameworks, important research findings, and any formulas or critical data points from the notes.

   When you finish, your output should be a carefully curated set of flashcards that thoroughly covers all the content from these 10 lectures, making it possible for a diligent student to score 100% by using these flashcards alone."

4. Academic Paper Analysis Tool
   Bit of a shameless plug here, but I'd be lying if I said it wasn't my most useful tool along with ChatGPT. You can read more about the tool here, or get insight into how it was (originally) designed here. Yes, I made the tool but I'm not begging for your money I promise (I am). All jokes aside, it has saved me a ridiculous amount of time and I feel I understand my readings so much better because of it... and those that use it all say the same.

5. Any Other Gems?

   Honestly, get creative, AI can help make most areas of your life easier and more efficient. During exam season, AI helps plan my weeks so I can work around uni, business and sports commitments. It helps me eat well with low effort when I'm busy with work and can't be bothered to cook. It helps me find academic papers for essays, especially with OpenAI's Deep Research capability now being availble to ChatGPT Plus users. And of course, it helps me massively with running a business.

Recommended Tools

1. ChatGPT

   • If you've made it this far, I don't really need to say much more about ChatGPT.

2. Claude

   • While I don’t personally use it, Claude is a useful alternative if GPT is overloaded or you want another perspective on a topic. It's models match OpenAIs (GPT) in performance across many tasks. Deepseek R1 is also a good alternative to the o1 model - I would use Deepseek R1 within Perplexity, as it is hosted in the US therefore bypassing Deepseeks censorship issues.

3. Perplexity

   • A free, research-oriented tool that excels at digging up academic sources, saving you from endless Google searches. They have recently integrated reasoning models into the AI powered search engine which makes it an incredible tool. I don't google anything anymore, I just use Perplexity.

4. Notion

   • A powerful platform for centralising all your study material. Its built-in AI features can also be used to query your notes or auto-summarise large documents.

5. The Academic Paper Analysis Tool

   • Last, but definitely not least, this tool is one of my proudest pieces of work. It has helped many students, myself included.

Integrating AI into your university studies isn't just about saving time; it's about transforming how effectively you learn, revise, and manage your workload. From crafting precise prompts to selecting the right GPT model for each academic task, leveraging AI tools thoughtfully can genuinely elevate your academic performance.

Remember, AI isn't a shortcut to skip the hard bits - it’s there to make them more manageable. Whether it’s rapidly producing high-quality lecture notes, custom flashcards, or analysing dense academic papers, the techniques outlined here can significantly streamline your revision process.

Experiment, refine your prompts, and don’t be afraid of trial and error - it’s how you'll find the optimal workflow that fits your study style. And when you do, you'll likely wonder how you ever managed without AI’s assistance.

If you found this helpful, or have your own AI-driven study methods, we’d love to hear your experiences, or any ideas you may have. Thanks for reading, and good luck with exams.