Transcription
There are millions of research papers being published every single year, and that number is growing by approximately 7% every single year. That means that very soon, the number of papers in your own discipline will double. How can you even keep up with the literature, not to mention get insights from other fields, especially if you're a busy professor preparing classes, supervising students, marking exams, and doing all that admin and faculty stuff that you have to do?
So, that's why in this video, I'm going to show you a simple three-step AI workflow that completely automates the whole literature review, saving you hundreds and hundreds of hours of your time and allowing you to ethically write exceptional literature reviews for Q1 journals, but also to discover really important, insightful new topics that will get you published in Q1 journals.
So, the first tool that we're going to use is Consensus. I like to think of Consensus as basically a research-backed version of Google. So rather than, you know, come up with complicated search strings, you just ask it questions, and Consensus gives you answers backed up by hundreds of research papers that it has access to. So, for example, recently I've been preparing some materials for published researchers in terms of what it actually means to have high-impact topics, what is a high-impact topic, and so on. So I can ask Consensus a question like this, and what it's going to do is read the hundreds of thousands, if not millions, of papers that it has access to and prepare a quick summary for me based on that research with references. So, the great thing in here is that it's not inventing stuff, and I can always look up what it's basing the information on in here, and then I get the reference list right in here. So, first of all, when you're diving into a new topic, just this summary on its own, right here, is worth its weight >> [music] >> in gold, really. Because this, you know, something that would have taken you months and months of work before, it's now done in minutes. And you've got a really detailed summary of what's going on in a specific field, in a specific topic, right?
So, then you obviously want to dive deeper into these papers because this is clearly not enough for a literature review for a Q1 paper. Before we dive to the second step and I show you how to find out more information, if you want to use Consensus, there's a link in the description and there's a discount code as well, in case you wanted to sign up for a paid version of Consensus if you're planning to use it regularly. [music]
Now, the second step is to start downloading all these papers, right? So, when you click on the paper, sometimes Consensus already has access to the full text, and it will take you, you know, to that paper, and you can just download it onto your computer. And the reason why we want to download them is that we want to give our next AI tool full PDFs to read so [music] we can actually get more in-depth information. And this next tool that we're going to use in this literature review AI workflow is Notebook LM. What I love about Notebook LM in comparison to, you know, Gemini, ChatGPT, Claude, is that it's [music] designed specifically, first of all, to keep all your files and charts private, so it never uses them to train AI models, it never shares them. And the second thing that I love about it, which is really important for the literature review, is that it doesn't hallucinate, ever. It always bases the answers on the knowledge base that you give it.
So, let me show you that second step in a little bit more detail. Once you've created your account, you create a new notebook, and I've created here for myself a notebook on high-impact research topics. Basically, researching, you know, what constitutes a high-impact research topic based on bibliometric research. What you do first is you add sources in here. You could add them from the web. If you're doing that, you definitely want to do deep research. But as scientists, we obviously want to base our conclusion and the literature review on published research papers, not just what is found on the web in general. In here, what you're going to do is you're going to upload all the files that you downloaded from Consensus from here, okay? And you're going to upload them to Notebook LM. I've already done that here. You can see I've got lots and lots of studies on this. And now I can just ask Notebook LM questions. So, let's say I want to know what actually constitutes a high-impact research topic, and I want examples from previous research. >> [music] >> And then what it's going to do is going to read all these sources to answer that question. By the way, [music] if you want such systems and workflows for using AI to publish more in Q1 journals, I've already built them for you on Published Researcher. I'll show you exactly how to use AI ethically, how to improve AI output, how to prompt engineer, and all of that. And I also give you custom-built AI research bots to help you publish in Q1 regularly. And if you're interested in that and all the other systems that we've developed, there's a free video somewhere in the description where you can watch a case study and book a free one-to-one consultation with you so we can help you to publish three to five Q1 papers >> [music] >> every single year.
Now, back to Notebook LM. So you can see what it does in here, right? We've got the first element, an atypical combination balance with conventionality, right? That's the first element of high-impact research topic. The true power of it is that then it gives us this hyperlink here, and I can see what it's basing the information on in here. So I can immediately judge whether I think this conclusion actually makes sense or not. And I can then go and view the source in here to start reading it to verify [snorts] that information. And the cool thing as well is that if you were to ask Notebook LM a question that, you know, it doesn't have information on because it's not contained in its knowledge base, then it's going to tell you, "I don't have the knowledge to answer this question. I don't have information on it. Instead, I can tell you a little bit more about something." Right? So, it won't hallucinate. That's the great thing. Okay?
So, you can then use it to research more deeply on specific aspects of that topic, right? Ask it specific questions. What you can also do, for example, if you're a busy lecturer, right? You can prepare slides based on these materials. And you can just like click here, and Notebook LM will prepare slides for you that are looking incredible. It will also prepare flashcards. It can prepare infographics. You can also prepare podcast or audio summaries, right? So, imagine like, you know, if you're really busy, you commute to work for an hour every single day, you might not have time to read the papers, but you can create audio summaries for yourself of all this literature that we've got uploaded here or of specific papers that you want to listen to in more depth, right?
So, this is the second part of the workflow. The first part, we use Consensus to get an overall synthesis of the literature, to get the relevant papers. The second part is to dive deeper in these papers and understand, you know, in more depth each aspect of the literature. And of course, we can iterate that process, right? So, maybe you've done 20 papers from Consensus, and you're starting to discover some more in-depth topics. Well, you go back to Consensus, and you ask questions on those in-depth topics. You get more literature. You plug it into Notebook, and you continue with it until you feel you're done. And just this is going to save you months and months of your time.
The third really important thing is then to use all that information to actually create something new, which is for example to write the literature review for your PhD thesis or for a Q1 research paper. And to do that, I think it's really good to create a custom gem or Gemini or what's on ChatGPT called custom GPT. So, it's basically like a customized, narrow version of AI that has specific instructions for one specific task and it knows exactly how to do that task, but nothing else. And this is really useful for repetitive tasks. So, you don't have to repeat that same instruction over and over again. And you can see I've got a lot of gems in here that I've created for myself. And we also have a lot of AI research bots on Published Researcher. But we're going to go to our AI literature review writer in here, to that bot. And what I'm going to do here is [music] plug in notebooks. Okay? So, Gemini directly connects with Notebook LM. And I'm going to choose the notebook that we are interested in, right? And then I'm just going to tell that we want the structure of the literature review based on the Notebook LM attached. And we're going to click enter.
Now, the way this bot is structured is that it's basically a sequential type of writing the literature review, so that you don't get the random answers from AI. Because the problem is, you know, or the misconception why people think AI gives you random answers is that, you know, AI just doesn't understand stuff. It's just not good enough. But the truth is that unless you provide AI with the right information, it's not going to give you the right output. [music] So, a good pattern to follow is what it's called in prompt engineering, flipped interaction. So, rather than, you know, you asking AI questions, AI asks you questions. Okay? So, for example, the first thing that AI is asking me is what type of paper are we actually writing here? An empirical, a review paper, or a theoretical paper? Because depending on that, the literature review will differ.
So, the next step in that sequential process is to provide AI with the scope, right? And the easiest way to do that is to give it the notebook that we've already created because it has all the papers. And the advantage of that is that, especially if you're on the paid version of Notebook LM, you can have up to 300 sources in one notebook. And once you've done it, it will also ask you for model papers. Like this is a crucial step as well because otherwise AI doesn't understand what a literature review looks like in your field. In the research papers written for that specific journal that you're submitting to. So this is a crucial next step as well. And then we also want the current status because maybe you're starting completely from scratch. Maybe you've already got some notes, or maybe you've already got something polished, right? That you just want AI to review, okay?
And once you've given AI all that, it's then going to generate a literature review for you. But actually, before it generates that literature review, based on everything we've done so far, what it does is it prepares a strategy for you. So it gives you a step-by-step guide to what a literature review would look like in your field for that specific journal with a lot of detail. And the cool thing as well is that it's referring to the knowledge that we've given it. And then it prepares this really nice synthesis of the guidelines with example context from model papers, the typical language that is used, right? And the target length. So the great thing about this is that then you can basically refine it a little bit, copy and paste it, and use it for any other future paper, saving yourself tons and tons of time. And after all of this, it's now generating the structure for the literature review. And we've got the main elements and then very detailed step-by-step elements in here as well. Just look at how detailed this is. And when you've done all these steps properly, right? This is almost a picture-perfect structure of a literature review because it's tailored to your specific field, to the model papers. It's got the knowledge, i.e., the papers that are in Notebook LM, so it's not hallucinating, it's actually using the knowledge from published research papers that we've given it to, and it knows the process of writing the literature review because this bot has been trained on our published research materials. And then obviously, you can start turning that structure into a written literature review. And if you want to know how to then write that literature review so that you tell a really coherent story, watch this next video where I show you exactly how to do that.