00:00Elicit AI is the granddaddy of AI tools for academia and research, but they have new things that you should know about
00:07So, if you head over to Elicit, this is what it looks like at the moment in 2025
00:11And these are the things you can do at the top
00:14You can generate a research report
00:16You've got systematic review and fine papers
00:18Now, depending on what amount of money you've paid them, they have different levels
00:23So, I was interested in something that I've not seen another tool do
00:27You'll see what I mean in a minute
00:29So, this is my account settings, but down here is the sort of like levels that they offer
00:33So, you've got basic and you get unlimited searching across millions of papers
00:39Absolutely love that
00:39And then you don't get everything else
00:42So, you don't get a systematic review, but you do get summarization up to four papers at once, chat with papers up to four papers, and then 20 PDFs per month
00:49So, it's a pretty generous free offer
00:52then sort of like $12 a month
00:55You get um a little bit more kind of flexibility in the number of activities you can do on elicit, but it was this that I was actually interested in
01:05It's $49 a month, but if you pay annually, it's $42 a month, but I was only testing it for this month
01:12Um you get unlimited tick, and it's this one here, systematic review workflow
01:17Now, I've been told that this is pretty powerful
01:20So stay around in this video because you'll see exactly what you get for this amount of money
01:26And it was pretty good
01:26It was pretty bloody good if you ask me
01:28So um yeah, let's go back into elicit
01:31So research report systematic review you don't get until you've paid the money, but you do get to find papers
01:37So the one thing I like to do whenever I head over to elicit is I'm normally searching for papers or I generate a little report just to kind of like understand a certain thing in the literature
01:47And so we can go here and generate a research report
01:52If we copy and paste this thing in what is the effectiveness of mindfulness-based interventions on reducing anxiety in university students and here it says great question
02:00Thanks elicit I know
02:02So then we can click go and then it will generate um you know a report for us
02:07What kind of answer do you need? Do I want fast the top 50 papers extract from up to 10 papers? Balance comprehensive and then we're going to use up to 10 paper extractions in my remaining amount of this billing
02:19So, I get 200 papers that it will review per month
02:22Um, so let's just go with the fast one because we're interested in a fast result and then it will go away and do this kind of like quick um analysis of the papers it finds
02:31So along the side here, you can see it's gathering papers, it's found 50 papers, it's screening papers, then it will extract the data and then it will generate the report
02:41And this is really what we're interested in
02:43We've asked it for the fast sort of version, so it shouldn't take too long
02:47If you go here, it's a systematic review
02:49If we go to um grab the same question, what is the effectiveness of mindfulness-based interventions in reducing anxiety in university students? Great question
02:58And then we got populate steps with suggestions based on research question
03:03And I highly recommend you keep that ticked because it's going to help sort of like improve the flow of your systematic review
03:09Now remember with a systematic review we look at all the data that's available all of the research papers and then we include or exclude certain papers based on whether or not they answer this research question and our selection uh criteria
03:25I do recommend you keep this because you'll see in a minute why that's important
03:28Okay
03:28Then we click next and it will send you through this kind of like structured process
03:32I really like it because it's the steps you would typically do when you're doing a systematic review
03:37So here it's saying I could select all 499 relevant suggestions from elicit
03:43I could select papers from my library
03:45I've got eight uploaded at the moment
03:47Or I can upload my own papers up to 500 papers
03:49So if you've already done the literature search, it's not wasted
03:54You can put it into illicit
03:56But I'm going to click here see suggestions
03:58And here's all of the suggestions
04:00And I'm just going to add the 499 papers
04:02It is a load of papers that it will include that it thinks may be able to answer this research question in my systematic review
04:12And so here we are
04:12Choose papers for screening
04:14Alysa has some initial suggestions
04:15Blah blah blah
04:15But I'm just going to say, yeah, I I I actually like all of them
04:19Love it
04:19I can delete these papers
04:20I can try again
04:22Um, but I'm going to define screening
04:24You can skip screening and just use all of these, but I like the screening process because that's what a systematic review is all about
04:30So define screening
04:32go in here and uh yeah this is where we sort of like ascent through the very sort of simple process of excluding or including papers into our systematic review
04:42Now down here you can see we've got suggested columns like what what age range it is, the data separation, the intervention duration, the intervention uh isolation or intervention duration I should say intervention type all of these things I can include or exclude and because it's I've clicked that suggest uh you know columns this is exactly what I get
05:02If I had that unchecked I wouldn't get this bit
05:07All right then then we can go to evaluate screening
05:09We can click the next button at the top and this is where you can see we are evaluating 499 papers
05:17At the moment it's looked at 56 and it reckons 56 could be included or uh 24 papers um are just sort of like aren't suitable for answering this question because it's sort of like in our um exclusion criteria or not in our inclusion criteria
05:32You get the idea
05:34We're now going through, I say we, the AI is now going through looking at all the papers and saying, is this good enough to include based on what I want to know? Yes or no? And you can see it's included this one, included this one, included this one
05:46It goes all the way down, all the way down
05:48And if we look down the bottom, you'll see it's excluding this
05:51All these colors, red, orange, blue, just mean like, yeah, this is an inclusion thing, but these ones don't quite make the cut
05:58So overall, this is an exclude
06:01And it will do that for all 499 papers
06:03Absolutely insane
06:08So, while it's doing that, you can see it's 40% complete
06:09Let's head back to that first question that we were interested in
06:14And it's extracted 70 data points and it's generating a report for me as we speak
06:18So, remember this is the simple fast report
06:20I love that it's generating this report in the background
06:24You can see that 2 minutes has elapsed at the moment, but it reckons it will take 5 minutes to finish
06:29So, it's still getting some citations
06:30It's sort of like giving us um an update as it's going
06:33Go grab yourself a cup of tea
06:35Why not? So, let's head back to the other one um where we're looking at the mindfulness
06:41We're 79% or 80% complete at the moment
06:44This would have taken in the past hours to read 499 papers and look to see whether they should be included or excluded from your systematic review would take hours and hours and hours
06:58And you can see that this is just going through it on its own
07:00And down here is the most important thing, score threshold
07:05So here for each paper, it gives you an inclusion score, whether or not it should sort of like be included or not
07:12We're finished with the 499 papers
07:15You can see this one has got a score of 4.9 out of five
07:17And that tells me that that is very very closely related to my research question
07:22and it satisfies all of the inclusion uh criteria or very close to it as you can see
07:29Then down here you can see they get less and less and less until all the way down the bottom we got 0.1
07:33And we can decide how many papers are included or excluded by sort of like moving this very simple slider
07:40So I only want things that are highly related
07:43So I'm going to go up to four and see or 3.9 and see what happens
07:47So in this case 107 papers are included
07:50If I increase it more, you can see that only one paper is included if I want it to be greater than 4.8
07:58So let's in uh decrease that so that we get 107 papers included in the report
08:05Excellent
08:05And then we define the uh extraction of data
08:07And here it says use suggested extraction columns or add your own
08:12So here it will go through all of the papers that we've just sort of like decided we want in our systematic review
08:18And now we can just uh yeah add custom columns
08:21We'll see in a moment it's suggesting other columns
08:25So here study design, randomization details, all of that
08:27And if we want to add custom columns, we can
08:29But you can see that it's running the extraction on all of that
08:33So each row is a paper and it's looking at study design, randomization details, participant demographics, all of those things that are selected down here
08:42And we can add our own
08:44Oh my god
08:44Step by step
08:44It's taking us through the process
08:47And we're not finished yet
08:48So stay around because we can run extraction
08:51Then we click run extraction and paid my hard-earned money
08:55I get 200 papers extracted uh a month
08:59Yes, a month
08:59So here I am about to extract 107 out of my 123 papers remaining in this billing cycle
09:05Yeah, no problem
09:07And then I can click here and extract papers
09:08Then it will go through and it will extract all the information from all of the papers
09:13And then after it's done that, I can say generate report
09:16So in my experience the generation of this report takes about 10 to 20 minutes depending on the number of references that you got inside
09:23Um I've done it for us so we don't need to wait for that
09:27But here we get the result
09:27So I've run this for us previously and I was asking this question
09:32Do retrieval practice techniques improve long-term memory retention compared to passive review in high school students? And it went away
09:41It did all of the steps
09:41As you can see, I was using uh 532 data points from 76 papers and then I generated a report and this is the report here
09:51So, it gives me an abstract
09:53You can see that it's all referenced um and we can go to methods
09:58If we click methods, we can see that it gives us that typical kind of um flow diagram of a systematic review
10:02You can see that it's got 498 papers from elicit
10:07I didn't add any manually
10:10And then it says based on um you know the stuff it was that many papers and then there were 40 uh included for the analysis and the data extraction and 422 were ignored
10:18And then we get screening, paper search, uh we get data extraction, study design, all of the bits that you would typically have to declare in a systematic review, it has got it here
10:31And it goes on and on and on
10:34It's got all of the different sort of like bits of data
10:36uh the retrieval format, the effect size, the control comparison, all of these things
10:43It went into the paper and extracted for us, which is why it takes so long, but you can go and have a cup of tea
10:47That's nice and easy, isn't it? Like that
10:49So, here we are at the end, all the references, and we can even spit it out in a PDF report that looks like this
10:57So, this is do retrieval practice techniques improved
10:59Exactly the same thing that I've just shown you, but it's in PDF format
11:03And you can see that it gives you a lot of information
11:05It tells you the inclusion exclusion criteria
11:07It gives you tables upon tables of all of the stuff that I wanted to know
11:13And obviously these columns you can put in uh custom ones when you're uh deciding what you want to include
11:19Um and it's just so much information
11:22You are saving yourself days if not weeks of work doing it this way
11:28Um, and overall it's just a really nice sort of like thoughtful process that you're forced to go through that you would typically go through anyway if you doing a systematic review
11:40And uh, let's head back to our very simple quick report
11:42Let's open up the PDF and have a look at that
11:44So this is the smaller faster report that I got
11:50You can see here we get data extraction again and we got results
11:52So, it's not as indepth, but it's still very very detailed
11:57And it gives you all of the different kind of conclusions
11:59We got outcomes and secondary outcomes
12:01We've got the references
12:03So, for a very very quick sort of like search and understanding of the literature, you could do something like this
12:09And it doesn't take very much money or time at all
12:15And I absolutely love that
12:15So, that is elicit
12:17And it goes on and on
12:21There are so many things here that you can do
12:22You can also just find papers
12:24This is how elicit started just by a simple um semantic search of the peer-reviewed literature, but it has become a really great AI tool that starts to do the hard work for you, which is what AI tools are all about
12:40So, here you can ask a research question and then you can explore the literature
12:44Um just like before you get a summary of the top papers here and then you get a row for each paper it's found and then you can go away and have a look at this by clicking on here source
12:52Um you can also filter it
12:55Here's your summary of the top eight papers
12:57Then you can add columns here
12:59So if you want to know about the limitations you can add the limitations and it will generate that
13:04You can see for up to eight papers at a time
13:06And then if you scroll down to the bottom you can add a new step
13:08So if you select some of these you can say you know what one paper selected add a new step
13:13create a table, summarize abstracts, chat with papers, add a new question
13:17So, it's a really great way of just getting a handle on what the research is saying, what you should do next, and a great workflow for really understanding the literature
13:27Absolutely love it
13:29So, here you can also do other things, upload and extract, you can summarize concepts
13:33So, if you've got a load of papers, you can look at um the the concepts across all of them
13:38And you can also chat with papers
13:40So you can just sort of upload a paper and then you can chat with that paper uh specifically so you can understand what's in it
13:47And by chat here we can add uh let's have a look we can add a summary so that we have a columns for summary for each paper and then if we select that paper here and then we can click add a new step chat with papers and then we can chat with the paper um use the full text which is not just the abstract and so we can say what are the main findings from this paper
14:15There we are
14:15And now we're chatting with all the papers we selected up here
14:18And it kicks out the results in a really nice simple way
14:21And I love that, you know, it yeah, just sort of spills out the information
14:25It's all referenced as well
14:27So you can click here, it will take you to the bit uh that it's referring to
14:31Overall, just elicit is becoming an absolute powerhouse
14:34It has never been easier to get an overview of a research field anymore, and Alysa is becoming the cornerstone of how people and researchers do that
14:43If you like this video, go check out this one where I talk about the free and easy tools that you should be using as a researcher
14:50You'll love it.