YouTube Video qEe5Zd-pRPo ![rw-book-cover](https://img.youtube.com/vi/qEe5Zd-pRPo/maxresdefault.jpg) ## Metadata - Author: [[YouTube]] - Full Title: YouTube Video qEe5Zd-pRPo - Category: #articles - Document Tags: [[fact]] [[research]] [[writing]] - Summary: YouTube Video qEe5Zd-pRPo - URL: https://www.youtube.com/watch?v=qEe5Zd-pRPo My mission in this video is to save you hours on your research. You see, Gemini AI is secretly becoming my favorite academic deep research tool. And here are five ways that you can use it to make your research easier and dare I say it, even fun. So, deep research tools have become a staple of the academic world. They are doing a lot of the heavy lifting for you. Essentially, instead of having millions of browser tabs open, you can now send out a tool to do deep research for you. And one of the first ways that you can use deep research is by using it for literature and scoping. So I do lit plus scope. And in this way, you can kind of like get a little bit of a map of like what research is out there, what is going on in the research field. And that is what I really like about it. So here I am on a Gemini advanced. I'm here on 2.5. This is the preview mode, but it is the most powerful thing I have tried so far in the large language model world. And you can see this is the prompt I put here for just scoping out a new research field based on the peer-reviewed research from the past 10 years. What are the major knowledge gaps or unresolved challenges? Organized findings by subtopics, blah blah blah. And then it kind of gave me this kind of research plan. and I I click start research and then I relax and then I go and do something else. I don't know, go to the tea room, go to uh do something that makes you happy. Um and ultimately this is what we ended up with down here. And we can click open and you can see that it's given me a really really nice overview of a research kind of field. And you can really use it for finding the first touch points for a research field, but also scoping out what the current sort of lay of the land is of a large research field. So here you can see it's given me a lot a lot of information. Scroll, scroll, scroll. It's got tables. It's got references and I can go down and click out to any of those. And uh yeah, here are all of the references at the end that are completely clickable. And then we've got show thoughts. This is how it sort of like went through its sort of like thinking process. Um, but ultimately this is just a really great way to start any research field and run this regularly just to see what's going on in your research field. Absolutely love it. That's the first way, but there are four more that are just as powerful. So the second way that you can use any sort of deep research tool is by writing. Now this is the most controversial kind of part of using any large language model but really we're using it to support our writing to justify what we've written to find references instead of saying now like oh I need to go and find some references to support this. This can do it for you. Now we have to be careful we're not cherrypicking the data just as if we were writing without AI. But this is a really great way of going getting references producing literature reviews and just doing the grunt work. Like can you imagine saying to people in the future that the way we used to write is by smashing our meat fingers onto keyboards onto plastic to create words. I think those days are behind us once academia really embraces this. In the future people will be using chat GPT and other large language models to write for them as a first draft. So let's have a look to see what's going on. So, if I go to my uh prompt here and I've got find recent peer-reviewed articles, policy documents, review papers that explain why research into scalable blah blah blah summarize the broader significance. So, this is really just a literature review. And this is what I want to use as like a first draft of um any sort of like, you know, thesis introduction even. So, it did an awesome job. Look at that. Oh my god, it's so long. It is so long that scrolling with my poor little finger. Oh, don't be sad, little finger. It's okay. We can just uh scroll scroll scroll all the way down to the end. But this is where I really like it. You can click up here to export to docs. And this is where I really like to kind of like start working with this file. You can see that it's uh really nicely formatted. We've got all of the little references in here. That's number four. It's a shame that they're not clickable and completely sort of um you know editable, but it's a great place to start. You can go in and put in your own sort of endote or mendlay or zotterero references here in place of them. It will take a bit of time, but you know, it's saved you a lot of the meat mashing into a keyboard time, so you might as well uh use that for referencing. Anyway, we can go all the way down and you can see it's got 38 pages and 42 references. In the past, this would have taken me days of reading, of researching, all of that sort of stuff. But it's done now for you. I know you don't like hearing done for you, but it's just the way this is going to go. We just need to wait for the institutions and the uh publishers really to come on board with allowing people to use large language models like this for their first draft. And uh I think it's inevitable. They just need to play catchup. So the third way that you can use deep research tools and Gemini AI is by critiquing. So let's have a look. Let's go argument. I can spell that better. Argument. So we want two opposing arguments. One, two. And we want to see if there is any disagreement in our current thinking compared to what's out there in the literature. You can use these deep research tools as a way of really making sure your arguments are robust by understanding what's out there in terms of opposing views. So, let's have a quick look at the uh sort of prompt that you could put in here. We've got what are the main critiques or alternative perspectives. That's the important thing. Alternative perspectives. So, alternative perspectives challenging the idea that vertical phase separation improves OPV devices. um and then summarize the core arguments blah blah blah and that's what we got. Once again, it goes away for like a few minutes and it produces this awesome document here. So, this is a really great way of making sure that your arguments are robust and using large language models in this way to really sort of like solidify your ideas, understand opposing arguments and viewpoints is I think what academia is all about. It's all about kind of disagreeing on some level. And now you can go away and see what the opposing arguments are in a super simple and non-confrontational, that's the word, confrontational way. This is just a really great way of increasing academic rigor. Try it for yourself. Now, there's one thing that really separates like mediocre scientists and researchers from amazing ones, and that's the ability to clearly communicate. And I'm not just talking about like speaking and giving presentations. I mean really convincing someone that what you are going to do will change the world and they should give you money. It's all about the money at the end of the day is it? So this is where we're at. We need uh communication. So let's just say coms has never been easier because you can now get money. Let's put loads of money there. That's what we want. Oh, look at all that ching ching ching. Um, ultimately we really want to use these tools for providing substance to what we're trying to sell to the world. And if we head over to a great prompt, here it is. Let's have a look. So here, this is the really the important bit of the prompt. Why this research is timely and aligns with global or national funding priorities. It's those sort of like niche and nuances that was really difficult to get in the past from any other large language model. But now with deep research tools like Gemini AI, which is just really doing a great job for academia and research, we can get that nuance and make sure that it's front and center of any argument and communication we put forward to get us money, grant money. We love that. And so, yeah, this is the full prompt. Identify current research trends, knowledge gaps and you know from the past 5 years. Like I said this is the most important thing timely and aligns with uh funding priorities and then obviously provides citations and compelling rationale. That's another important bit. A compelling rationale for new projects in this area. You can see that it's done a really great job. has given me an executive summary and uh yeah, this is a great first draft of exactly sort of like the information that a funding body would need to know about about your research field to make sure that you stay relevant and you stay sort of like um cashed up. That's it really. the more and more research progresses, we need to make sure that you are actually sort of like looking at adjacent fields and looking for that interdisiplinary kind of um approach to your research. It has never ever been easier to now look across over the fence, over the neighbor's fence. Oh, hello neighbor. And be like, "Oh, what you doing over there? Oh, that's nice. Why are you in your underwear?" Anyway, I that was a true story actually. Once I looked over to my neighbors and they were in their underwear building a cat cage. Interesting. Hey, the world. Um anyway, so um you can now look across to your neighboring sort of field and see what they're doing over in their back garden to see if you can steal some ideas from them and also maybe collaborate with them in the future. It's never been easier to find that intersection between your field and someone else's field. So here let let's go down and for uh completion sake let's have a look for finding let's have a look uh let's have a what do we call this uh collab co oh go again co no okay okay collab there we are collab and that is about finding that overlap with someone else's research field and you're like you know what I like what you're doing over there mate give me some of that and so now we can head over and see a great prompt and here it is. Let's have a look. Uh it says how have researchers in adjacent fields like biomedical imaging or nanofluidics and you can put your own adjacent fields there. If you don't know what adjacent fields are, you can just ask chat GPT or another large language model like what are the adjacent fields to this thing and then uh yeah you can put in um the summarize any transferable techniques or insights and site relevant s studies and this has been going on for ages. It's researching 142 websites, but I can assure you that it will give me a good result. It's been going for like half an hour at this point. I just wanted to film this video. Uh, so I was like, I got to do this anyway. So, it's going away. It's doing a great great job. But stay around because there's a bonus tip that I want you to know next. All right. We've got a bonus tip. Look at this [Music] bonus. Okay. Yeah, you get the idea. Bonus tip. Bonus tip. Okay. So, over here, the one thing about Gemini that I've not used properly yet, but I I did recently and I want to show you it is Canvas. Canvas is a really great tool. It's h allows you to interact with the what it generates. And instead of sort of like regenerating everything when you give it a new prompt, you can work on individual sections. And I was using it for academia research and this is what I found. So, this isn't deep research, but you can see here that I just sort of like did a typical literature review scoping type thing. And I just said create a structured overview for the key things, blah blah blah. And what it did is it gave me this. It gave me this thing which allowed me to sort of like work in it in Canvas. And here I was able to kind of like um uh edit and do all of the things I would do typically in Docs. Um, here I can export to Docs, but because I haven't clicked uh canvas, you get it. I clicked away from it. It doesn't work that way anymore. But down here, these three buttons are really awesome. Change length, change tone, and suggest edits. So, the one thing I really like about this is you can go in, for example, in this bit there were no references. So, I can go in and just say, "Hey, give me references for this bit." And the references, by the way, some of them were great and some of them weren't so great. If you go scroll down to the bottom, like I was like, "Yeah, that one's real. That one's real. That one's real. That one's real." And then all of a sudden I got down to 15. And even though this paper exists, it's not under this name. So there is a little bit of hallucination, like a little bit of blurring of the facts. The paper by these people with these sort of like uh numbers, you know, issue 17, whatever. Five, I forget what five means. these pages. It exists, but it's not under this name. So, we just have to be super careful when we're using references um that aren't sort of like specifically linked out. But, nonetheless, using Canvas is a great way now of working with Gemini. And I think you should give it a go. If you love deep research, check out this one where I talk about how academics are secretly using deep research to make their lives easier., YouTube Video qEe5Zd-pRPo, YouTube, 01kac36v52k4z8caravvrjj2tm.