Speaker 1: I may have found something that's better for research than ChatGPT, it's called Perplexity. Perplexity.ai and the reason I think it may be better is because it has access to a load of different sources that makes your information completely up to date. So you can see here it's got a very similar layout to ChatGPT but under here we can see that it draws from the internet, from academic sources, from WolframAlpha, from YouTube, Reddit and News. Okay on the front of it it doesn't look too different apart from these buttons. We've got this drop down menu here which is quick and concise answers and enhanced. Enhanced is what they use with GPT-4 so it should be just as powerful as ChatGPT plus the stuff that you pay for. This however is free. So let's go on to enhanced. Let's ask it for the latest papers on organic photovoltaic devices. Let's see what it comes up with. Now the one thing I like about this is not only does it give you the result but it also gives you related questions and references. In ChatGPT-4 there were some potentially made-up references definitely in GPT-3 model but in 4 there were some blurry edges that we weren't quite sure whether or not they were real, if they existed or if the DOI was actually the right DOI. But here we have got not only the latest papers but we've also got the references from Semantic Scholar and whatever this one is. Let's click on it and have a look. So here this is from the National Library of Medicine. Let's try again. So give me papers from the last year in the field of organic photovoltaics. If this is truly going to be useful for us it needs to be able to access the recent literature, something that ChatGPT-4 does not do. So we can see we've got 2021, let's say 2017, 2021. Please note the second paper is 2017 but it provides variable context. Okay so at least it's trying to help us. There's a couple of things I like about this as well as you can go back and view the sources and you can add more sources but also you can edit your query. If it's not quite what you wanted you can go back and have a look to see if you can change the query without having to start a new stream or go underneath the previously wrongly answered question. So I do like that sort of stuff. The great thing about Perplexity is that it has got an iPhone application and it's got something coming for Android at the moment. For anyone accessing ChatGPT on their phones it is a little bit clunky because you have to use the web interface but with a dedicated app Perplexity.ai may just make it that little bit more seamless as you're in the lab doing research and you just want to look something up quickly. You don't have to plug in information but it doesn't read PDFs as well as I was hoping. There is a browser extension but you can see that on one of my papers if I click the browser extension and say summarize it doesn't really do anything. The search result is empty and no information is able to provide an answer so it means that we still have to copy and paste information from the papers into the model to get those responses. There are sort of different ways you can get it with ChatPDF and also HeyGPT which is my personal favorite, something I've paid for, but you can see that there's still a few areas of blind spots if you want for academics. But let's do a side-by-side comparison of Perplexity and ChatGPT. I've got here a simple prompt and it says act as a scientist and create an outline for a literature review about organic photovoltaic devices. That's my area but feel free to try it with your own field. The first thing I've noticed is that ChatGPT has a much more in-depth and long-form content generation for this. It is something that's invaluable if you just want to take the grunt work away from your mind so that you can actually focus on the details and here you can see it's produced quite an in-depth summary that we could go through and look at whereas over here it's much shorter, I would say probably about a third of what ChatGPT put out, but importantly it's given me references down the bottom. Now how useful they'll be are right now I'm not quite sure, but you can see that here it's actually tried its best to kind of plug them in. I'm not sure that the references are completely useful at the moment from Perplexity AI but it's useful to be able to see where it's getting its information from, something that's just a little bit of a black box from ChatGPT 4. I think this is a win to ChatGPT. Let's try another one. Let's say you don't understand a concept, you just need someone to explain it to you. Let's have a look to see what Perplexity does if I just ask it what upconversion is for organic photovoltaic devices and I'll do exactly the same thing with GPT 4 in ChatGPT Plus. So many GPTs. Okay let's have a look to see what happens. Once again ChatGPT Plus is giving us a much longer in-depth response, something that may be useful. However, Perplexity.ai has given us actual references for us to go and have a look at where they're getting the information from. So I think once again it could be a little bit of like a win for ChatGPT because it just gives me the information but it's also kind of a win for Perplexity because it is providing its references. Now I think as a scientist I would actually still prefer ChatGPT but Perplexity has its place when you want those references. Overall Perplexity.ai could be something that is invaluable to a normal sort of research environment. Getting academic references is just such a pain going to find them whereas Perplexity is trying to make that a little bit easier. Now the thing about ChatGPT is that at the moment it's just completely text. It doesn't search the internet although it was teased at one point and it does not read PDFs but there are people building layers on top of ChatGPT that allow you to access those things. One of my favorites is HeyGPT and I've talked about how I've bought this in the past. It's fantastic. So over here you can chat with websites, chat with files and it's got Wolfram Alpha coming soon. So this is a one-off payment and you do have to use your own API key and it can get very expensive. So here I've been given access to ChatGPT 4 as an API which means I can run it in something like HeyGPT but most people will only be able to access the 3.5 which is much less expensive and much more affordable. You'll see that if I go back to chatting with my documents. So here you can see I'm chatting with one of my files and that file is one of my papers. Planar, silver, nanowire, carbon, nanotube, blah blah blah. You can see that if I use that, chat with files, this one, apply. So I've asked it to summarize this paper for me in five dot points. ChatGPT 4 is much slower so it's writing out very slow but you'll see that it's actually quite an expensive way of talking to documents. You can do it much more cost effectively using ChatGPT 3.5. Now we can see it's only cost us about eight cents for that query however if you do this time and time again it soon adds up. You can save money by using ChatGPT 3.5 model but overall I prefer GPT 4 because the answers are just so much better and it's worth the little bit of extra expense to get really good answers and that's how I can use HeyGPT for accessing different PDF files and getting the information and chatting with those documents. Now I do know that ChatGPT is working on plugins. Once they are released to the world I really feel like these other companies like Perplexity AI and also other companies that are building on top of ChatGPT like HeyGPT are going to struggle. They're going to have to now compete with a whole new functionality that will just open up ChatGPT to so many different applications, the internet and more. So there we have it there's everything I think you need to know about the current state of AI tools for research. There are competitors to ChatGPT that are getting better and better however people are building on top of ChatGPT and including things like reading PDFs, websites, asking Google and I think that is going to be such an awesome thing for researchers just to make things so much more sort of smooth when you just want to grab that reference or get something from a PDF that you don't want to read. All of these new capabilities are going to really benefit researchers and you need to keep on top of them. So let me know in the comments what you would add and also remember there are more ways that you can engage with me. The first way is to sign up to my newsletter. Head over to andrewstapleton.com.au forward slash newsletter. The link is in the description and when you sign up you'll get five emails over about two weeks. Everything from the tools I've used, the podcast I've been on, how to write the perfect abstract and more. It's exclusive content available for free so go sign up now and also remember to head over to academiainsider.com. That's my new project where I've got my ebooks and my resource packs as well as my blog and the Insider Forum. It's all going off over there so remember to go check it out because it's all there to make sure your research and your PhD works for you. Alright then, I'll see you in the next video.