Speaker 1: When you first go to Perplexity, it looks like this. This is perplexity.ai and first of all, you can see it's very, very simply laid out. Where knowledge begins. Come on now, don't big note yourself. And you've got this, ask anything. You can see you've got options down here of maybe things you'd be interested in, but we're interested in academic research, they're no good to us. So, this is a very, very familiar layout for those that are used to something like ChatGPT. And asking anything in here is where it all starts. Now, there's a few things you need to know about even in here. You've got CoPilot. What is CoPilot? CoPilot is where you have this kind of companion, which will not only just sort of like work out what you actually want, but it will ask clarifying questions to make sure you get the perfect answer every single time. So, here I've always put CoPilot on. Because I'm on pro, I'm a pro. I don't know why they're always called pro. I want to be called something awesome, like Mr. Wizard. That's what I'd pay for, I'd pay more for that. But when you're on pro, you get like 600 CoPilot searches a day, but on free, you get five, which is probably enough for most of the things people want to be doing daily. And then down here, you've got focus. If you click on focus, you've got all. You've got academic, you've got writing, you've got Wolfram Alpha, you've got YouTube, you've got Reddit. So, clearly, this is the one we'd be interested in, academic. But I quite often just leave it on all, because in the early stages of using this for research, I do actually just want to see what's out there. Later on, when I'm looking for specific literature, and maybe it's not giving me what I want, I can say, you know what, just look in the academic world. And then, of course, we can attach stuff. There's stuff coming up about attachments later on, which really, really blew me away, so stay tuned. And you ask a question. So, the first question I asked it was up here. I am starting a PhD, and asking my supervisor questions is still a little daunting, and they smell like coffee. Can you please give me review papers about transparent electrodes from the past couple of years to get me started? So, this is where Copilot popped up. It asked me a question. It said, which field are you interested in? Is it physics? Is it chemistry? Is it material science? And I said, I want chemistry and material science. And then it started searching the web for these sorts of questions. Now, it was searching the web because I had it on all rather than academic. I'll show you what academic does in a little bit. But then it found down here, you can see it found 19 sources, and then it gives you your answer. The answer is always very nicely formatted. I do like the fact it kind of, you know, puts it in this numbered list, and you get these little pop-outs here where you can click on it and you go to the reference. Now, I asked it for review papers, and so here you can see that it is a review, and it was from 2020. Hmm, close enough, but I feel like I can get a little bit sort of like later in terms of the review papers I would want to read. And then I can click through this one. Okay, transparent electrodes for organic optoelectronic devices. I can click that one. And to be honest with you, I have been relatively impressed by the sorts of papers it's managed to find. Now, this one is from 2014, still a little bit too old, if I'm gonna be absolutely honest, especially for this research field. But I think I can ask a follow-up question and just say, you know what, get papers from the past, you know, few years. This was quite a broad search. I did say from the past couple of years. It didn't really catch onto that. But remember, this is always a conversation, so if it's not giving you what you want, you can just say, hey, now give me this and instruct it to do what you actually want it to do. But it gets interesting when we want to look for specific papers. I used this prompt. I'm a postdoc researcher and I'm stuck on the postdoc treadmill. My supervisor is away on a sabbatical, which we all know means is a holiday that is tax deductible. Give me five and I said recent papers on nanomaterials for transparent electrodes that I can use to impress them when they get back. So this is where Copilot once again jumped in and it wanted me to sort of tell it what I was interested in. Here I chose performance and it was really easy. I just selected between three options and I said I want to know about performance. And then based on that, it searched the web and it went away and I was interested to see whether or not actually I lied before. I'm a big fat liar. This was all, this was using all. But because I was asking for papers, it did only actually provide me with really sort of like academic responses, which is really great, which is why quite often I just leave this on all. I don't want to limit it to say, you know, only look at academic papers. But you can do that if you feel like it's giving you too broad of a result. So here once again, we get this numbered list and I was sort of very happy to see that for example, this is 2023. Brilliant. So it has actually delivered on what I wanted, which was the recent papers. So you can ask it for recent papers and it does deliver, which is really fantastic. I'm amazed at the number of things that actually sort of able to sort of bubble to the surface just with one simple, easy prompt. So here this is from 2022. So yes, this is exactly what I would want to know if I was in this field right now. Really great and really powerful. Well done perplexity. And look, this is where it gets really powerful. Go down to settings and these are the things that people quite often ask me on this channel. They say, what about data? If I upload something, is it going to become part of their training model? Here you can turn it off, which I really, really like. And then down here, we've got all of the different ways that the AI model works. When you've got pro, you can change between different settings, which I really like. So here for AI model, we've got default, we've got experimental, we've got GPT-4, latest model by OpenAI, we've got Claude. So for one payment, you can get access to all of these models. Instead of just using chat GPT, you've now got access to a range of different other models, which is just brilliant. Image generation, if that's important for you, you can use default or DALI. And it comes with a pro discord, pro support, all that sort of stuff. But those are the most important things in the settings that people are often worried about. Really great, well done perplexity. But it gets even more powerful. I was amazed at what this was able to do with images. Just a couple of months back, we were amazed at chat GPT's vision. Now, vision has come to perplexity. Let's check it out. I was amazed at what it could do. Here, I've got a simple prompt. I am reading a paper and I don't understand this image. I don't wanna spend too much time understanding it because it's a little boring. Can you explain it to me? Once again, Copilot popped up. It asked me an understanding question, which I can't even remember what it was. But I clicked on through and I said, look at this image. It was like a simple image. You see it's got a little bit of small text. It's got arrows. It's got lots of really confusing components. If it can pick out the process in this, it could do it for almost any schematic that you generate or you find online. Look, there's an arrow going backwards. There's curly arrows. Oh my God, it's so confusing, even for a human. How does this deal with it? Well, so it says, the image you've provided appears to be a schematic representation of a process for fabricating a type of composite material likely involving carbon nanotubes and a polymer substrate. It even picks up later silver nanowires. And it says possibly silver nanowires in IPA. And it picks out the solvent isopropyl alcohol as well. All of that from this image where it's not written out at all. PEN plus epoxy, PEN. Like it is just, just incredible. So it knows that we have to prepare a nanomaterial suspension, yes. It knows that we have to filtrate and format the nanoparticle film. Then it knows that it's a transfer process. Then it knows there's an application of a polymer matrix. It just knows, it just knows. That's amazing to me. And if you were to upload the figure caption alongside this, I'm sure that it would just be able to explain almost anything. And then you can ask follow-up questions like how do I do this myself? What are the most important things that I should know about this process? All of those little things just will help you understand the paper from figures. Just incredible. But it gets even better. You can currently upload four images at a time. And I said, these are images from a peer-reviewed paper that I'm writing. It's currently in draft stage and I hate it with my entire being. Mm. We've all been there. Can you help me create a story and order from these figures? So when you're writing a peer-reviewed paper or even like a chapter draft that you're gonna give to your supervisor, you want to create stories. Stories are what attract people. And if you're not sure where to start, this is a great way of doing it. Put in the main figures. So here I've got that figure. I've got this figure. And they can be a little bit confusing because like I said, I've not given it any context other than these figures. They've got no figure captions, just the numbers and the stuff it can make out from this image. So overall, it says, here we are. To create a coherent story and order from the figures, we can construct a narrative around the development, characterization, and performance, and evaluation of composite materials. It consists of silver nanowires and single-walled carbon nanotubes. I did not tell it that that's what I was doing, but for some reason, somewhere in here, it has, oh, here we are, silver nanowires. And then down here, it's got silver nanowire loading. And then here, it's got single-walled carbon nanotubes. So it knows it from these little sort of like tips and little clues I give it in the figures, which is just incredible, something that's never, ever been able to be done before online with AI, just incredible. ChatGPT does something similar. I would argue that it's a little bit better, but still, this is very powerful, and you can go on to ask more questions. So sheet resistance versus silver nanowire loading, microscopy, microscopy and spectroscopy analysis, and mechanical flexibility test. So you'll notice that it actually didn't do these in order. It actually put this one at the end. That's not the order I uploaded it. So it is making decisions about the story it wants to tell. And then down here, it does tell you, like, we should suggest directions for future research. We should have a conclusion. And then it says here, by following this structure, you can create a logical flow that guides the reader through the development and evaluation of silver nanowires and single-walled carbon nanotube composite material, culminating in a clear understanding of its properties and potential applications. That sounds like the perfect paper for me to be writing. If you don't want these images to go to its training model, you can turn that off. Absolutely brilliant. Something that ChatGPT doesn't give you if that is important to you. And another way you can use perplexity for research is this next thing, which is super powerful. Let's check it out. Reading papers is a massive pain in the ass. So here, I said, this is a paper, and I uploaded it as a PDF by just adding it here. So you can say, attach images, text, or PDF. So I added one of my papers. And I said, this is a paper that I'm reading as part of my PhD. If I read another one, I'm going to throw my computer out of the window. Can you summarize the key points from this paper so I don't have to read it all? Once again, Copilot popped up. Bing. Or no, that's a different brand, Copilot Perplexity. Ooh, perplex, I guess it would say. Here it says, understanding the question, then it reads the file, and then it searches the web because it knows what I want. And then, because it had the title, it bound it online in a number of places, and then it gives me an answer. A really, really detailed answer, a really powerful answer, super easy, because I think it uses information that it found online and from the PDF itself. And you can see that it's got objective, methodology, performance, application in organic photovoltaic devices, advantages over ITO, and conclusion. So I can go on and ask it stuff. What is the main focus of the paper? I could ask it what the limitations is, what the next sort of step should be, if I should continue this line of inquiry, for example. All of those things from a single paper and just a click of a button and a simple prompt, really good. And here's another little cheeky thing that it provided me that I was very pleased with. For some reason, it's found a YouTube video from one of my old collaborators who's now, I'll turn the volume on, who is now a associate professor at QUT. I'm very proud of her. Well done, Sonia. But here she is doing all of stuff. So it actually sort of like looked at the collaborators and brought her up as well. So I know where she is at the moment, which is like another little sneaky tidbit that I didn't even know it would do, something that ChatGPT doesn't do. So I am very, very impressed with Perplexity. I'm very, very impressed with what it can do. And if you want to know more about using ChatGPT or AI, go check out this video where I talk about the epic ChatGPT prompts for research and science. Go check it out because it's an awesome watch. I'll see you over there.