Speaker 1: Here are 10 really perfect prompts for you to use if you are in academia or research. You can use these in ChatGPT, Perplexity, Claude, wherever you want to use it. And the first one is this. As an academic paper writing assistant, your task is to help users structure their scholarly paper efficiently. So this is about making a structure from which you can build upon. Now, when we are faced with like a blank piece of paper or a blank Word document, it can be very confronting to have to come up with words. This is a way of creating that. So here I say, create an outline about a peer-reviewed paper on, and then you can insert your own thing here. And then it says, here is a structured outline. So you can see it gives you the title, the abstract, and it also gives you these bullet points of things you can actually include. Background and literature review, material requirements, blah, blah, blah, it goes on and on and on. And this is how you can start strong no matter what you are writing. The more information you can give it upfront about what you are writing and what you want it to become, the better. So there's the first one, and importantly with all of these, the first thing is the context. Context, where you have to say to the GPT, okay, this is essentially who you are and what you are. That is how you are going to help me. You'll see that throughout all of these prompts. The next one is particularly powerful. So you'll see that I uploaded these figures. This is on chat GPT four, so this is where I have paid money, my hard-earned internet money, to use this tool. And so I've uploaded five figures, and then I've used this. As an academic paper writing assistant, that's to them, guide users leveraging figures to create structures in their research paper. Take the figures attached and create a narrative for this paper. And you can see that it has done exactly that. It does know these figures. Look, it's amazing. It gets all of this information from inside these figures, and it starts saying, well, here's a potential narrative. If you don't know if you've got a complete narrative for a peer-reviewed paper, put the figures, which is where I always start, into something like chat GPT vision, and it will tell you whether or not you've got a complete actual story that you can tell. You'll be surprised, actually, at the game of academia. You can tell research stories and create papers far sooner than a lot of people really realize. So use this and this prompt with the figures that you've created to see if you can write a paper right now, start today. Coming up with titles is a massive pain in the ass because you want it to be attractive enough. You want it to show the novelty so you can ask it to help you. Here we are. We've got this one. As a title evaluation assistant for academic papers, your task is to assess the effectiveness of a paper, blah, blah, blah. When a user submits a title, begin by evaluating the clarity relevant to content, and importantly, it gives you a number feedback. So it says rate each of these criteria on a scale of one to 2.5, leading to a total score of 10. Provide feedback for each criterion and highlighting strengths and suggesting all of that sort of stuff. So I put in my number one cited paper, and it's this one, the accurate thickness measurement of graphene, and you can see the clarity. You can't get much better than that. It is succinct and clear. Originality, maybe not so good down here, 1.5. So if there's loads of papers out there with this title, something very similar, it's hard to stand out. That's why people normally inject novelty into their titles. And then ability to engage the intended audience, another 1.5. So it tells me that if I'm struggling to get this paper noticed, or I want it to be even more impactful, I could include these sort of things. And you can even go on and say, provide some examples of paper titles that improve on this one. It will give you sort of like five examples, and you can pick and choose the parts of the titles that you like and mash it all together into a mega title. Love that. Here's the next one. Something that ChatGPT and a lot of large language models do incredibly well is analyze text. And you can tell it to look at your potential peer-reviewed papers that you're submitting and give you feedback. So here we got a prompt that starts with, as an academic peer review assistant, your role is to provide thorough and constructive review of submitted academic papers. And so I tell it to go through all of this. And by the way, all of these prompts are in either the comment, if I can fit it, or it's on a document linked that you can get. So you can copy and paste them into any large language model that you are using. But I want it to check for all of these things, which is great. And then at the end, I have to include this. Wait for me to give you the information. Because if you don't do that, sometimes ChatGPT and other large language models just start trying to please you. Like, oh, please, sir, please love me. And they just give you all of the wrong information. But now you just say, wait for me to give you the information. And it says here, absolutely, I'm ready to help you with the peer-reviewed process. And then here I've got, this is the introduction to a paper I'm writing. And I put it in. And then as you can see, it says down here, content accuracy and relevance, argumentation, yes, and analysis, methodology. And it also gives me all of these suggestions for improvement, really great. And you can use this, go backwards and forwards. And it's the first touchpoint to a paper before you send it to the editor or even your supervisor for checking. Send it through something like this so you can just avoid that long wait of your supervisor just having it on their desk for months. And you're going like, come on, just read the bloody paper. Now you can skip that step. Use this. Working with text that you have already created is a strength of large language models. So you can use prompts to extract stuff, to manipulate things. Here are my three favorites. The first one is this. You are an expert research assistant and you always start with context of how you want it to act because it just sits there and it goes, oh, yes, I'm a perfect research assistant, brilliant. And then it says, from the submitted peer-reviewed paper text, please create 10 keywords that best represent the paper. So when you're submitting your paper for peer review, quite often they ask you for all of this extra information, keywords, summaries, and you just need to go in here and just ask it to use the paper you provided to extract the most important keywords. It's super easy, saves that brainpower for something actually useful rather than those silly keyword lists. Another one is this. You're an expert research assistant. Again, create a short summary of this peer-reviewed paper. And you can swap that out to say, create a layman summary or whatever the journal is asking you to fill in a little box. You can just say, create a short summary of this peer-reviewed paper. Stop using your valuable academic time to reinvent text you've already written. That's what large language models absolutely excel at. And the last one in this vein is a two-stepper where we are saying, create an abstract for this research paper. So you're an expert research assistant. Again, create an abstract for this peer-reviewed paper. And then I would copy that into ChatGPT and I'd allow it to come up with an abstract that it thinks is suitable. But the problem is, abstracts aren't all the same across multiple fields in academia. So you can then back it up. The second prompt I would use is use these examples from other papers to help refine the structure. And then I'd put in example one from the journal that I'm targeting, sample two from the journal that I'm targeting. And then together, it will actually create a much stronger, more coherent abstract. And this two-part process really helps hone that perfect abstract for your paper. When you enter academia, you realize that every time you need to do something, you need to send an email. So I created this prompt which just allows me to create quick emails in the appropriate tone. And it's like an email generator. And the prompt starts with how it should act as an academic email writing assistant. Your role is to help users craft professional and effective emails to their principal supervisors in academia. Begin by guiding the user to clearly, and then it's got all of the stuff it should do. So down here, the last thing I say is, ask me the questions you need to write this email. Because otherwise, it just starts giving you advice. So it says, sure, I can craft the effective email. Let's start by clarifying a few key details. And it's got four questions it needs me to answer. Purpose of email, relationship with the supervisor, specific details needed, and desired outcome. And I put it down here. I just copied and pasted this. And, you know, purpose of email, comma, they've been looking, oh, the, the've been looking at my thesis for the last three months and I need it back now. Your relationship with the supervisor, intense relationship. And then specific details, none, desired outcome, I get my thesis back with corrections. And then as you can see here, it gives me subject lines, it gives me how to start, the introduction, the body, and the conclusion. It's just a really great way, if you're double guessing about how you should speak to a principal supervisor or, you know, an academic that you need to send an email to, this email generator really does work. And, you know, it really does help that little bit of anxiety. You don't want to hassle people, but sometimes you do just need to get in their email inbox. This is how you do it, love it. The last one I think you should know about is this. It's a simplification assistant that I've created this little tiny prompt for. Absolutely love it. And it says here, as an academic simplification assistant, your task is to help users explain complex academic papers in a manner accessible to a 14-year-old audience. Now, where did I get this? 14-year-old, like an educated 14-year-old, is really where I think a lot of really great science communication happens. And so you can reduce that number if you want, but for me, that's the sweet spot, where it's detailed enough that it's not like babyfying the language, but it's not so complicated that it starts getting into, you know, really complicated A-level or above type things. So here it says, begin by advising the user to first understand the core concepts, and then once again, I've got this. Wait for me to give you the content I need explaining. And then it says that. So it's a great plan. Once you provide the context, so here I put in an abstract from one of my papers, and you can see down here, ChatGPT kicked out this. Why use water-based polymer nanoparticles? It gives me the understanding. Problems with older methods, what's new in this paper, conclusions. So this little tiny thing just helps me understand the paper properly. And if you're in academia reading loads of papers all the time, I highly recommend you use something like this. Put it into ChatGPT or your favorite large language model, and it will help you understand without having to use all that brain power. You can put that towards something useful, like generating your own data and finishing your goddamn PhD. Now if you like the prompts in this video, I highly recommend that you save them to something like TextBlaze. TextBlaze is perfect for creating shortcuts so you don't need to type out or copy and paste all these things all the time. This is mine, and I use it all the time. This is ChatGPT folder, and this is all the things I regularly do in ChatGPT. I'm learning Persian, I have to put that in Anki deck, then I've got descriptions, I've got lessons, I've got shorts, I've got all of these things I regularly do, and I need ChatGPT to have that prompt, initial prompt, and I'm fed up of copying and pasting it and typing it out. Use TextBlaze, super easy and quick. If you like this video, go check out this one where I talk about the epic ChatGPT prompts for research, you won't regret it.