How to Use AI Responsibly in Academic Research (Full Transcript)

[00:00:00] Speaker 1: If you have questions around the use of artificial intelligence when it comes to evaluations and research purposes, then you are at the right place. My name is Anne-Marie Brown, and I'm an evaluator with over two decades of experience. And I'm joined by my colleague. He's the head of the Center for Research Methods, Dr. Philip Adu.

[00:00:25] Speaker 2: Hello, everyone.

[00:00:27] Speaker 1: Welcome. Now, Dr. Adu is perfect to answer your questions when it comes to using AI for your academic research purposes because Dr. Adu was one of the first persons I knew who was publicly using AI long before chat GPT even became mainstream, right? So the next 30 minutes, this is how it's going to go. It's all about answering your questions. So put them in the chat, and we'll try to answer as many questions as we can get in within 30 to 40 minutes. Dr. Adu, are you ready for the challenge?

[00:01:11] Speaker 2: Yes, I'll try to answer all the questions that I have. Yes.

[00:01:17] Speaker 1: All right. So let's go. First out the gate, first question. Which AI tool or model do you recommend for research purposes?

[00:01:28] Speaker 2: Okay, so I'm going to recommend what I have to use. And so I use chat GPT, especially GPT4, the paid version. I use it to analyze my qualitative data. And also for the paid version, you can create your own custom GPT. So that will do a specific job for you. We call it AI agent, right? And also I use cloud. Cloud AI is very good. The reason why cloud AI is good is that it's like it doesn't hallucinate. I hope I mentioned it right. It doesn't give you wrong information all the time. It's, you know, most of the time, you don't have to give a lot of context to get what you want. So cloud AI is very good for researchers, especially when you are trying to gather ideas and also think about maybe how to analyze your data. You can upload your transcript and then ask a system question so that you can get some answers to help you to analyze your qualitative data. Publicity is also very good because it gives you the source. You know, when you ask a system a question, it can give you a response and also provide you a source that you can click and learn more, right? So it also address, you know, the system giving you wrong information because you're always going to get a link to really go to the source and learn more about it. So these are the, you know, the AI tools that I personally use. And there are other ones, but these are the main ones that I use for my research.

[00:03:02] Speaker 1: All right. So to summarize, it's ChatGPT, cloud.ai, and perplexity.ai. These are your three recommended AI models for your academic purposes. And I also recommend them for evaluations as well. All right. Straight into the chat, we see a question here. Dr. Adum, please, what are the limitations of using AI tools in writing literature review?

[00:03:32] Speaker 2: Okay. So one thing is that I always say that we don't use, for academic purpose, you don't use AI to write things for you. You use AI as a tool to assist you to write what you have to write. Right. So I think that the reason why you are doing literature review also to get the skill in writing, scholarly skill, but allowing AI to do things for you is not the best way. So what I recommend is that you can get ideas from the system. You can use ChatGPT. I have this topic. Can you give me ideas about how to write? What are the subtopics related to this? And then you get ideas and then work with that and write it yourself and use AI to edit. If you have perplexity.ai. No, not perplexity. Pipapao can be used to edit your document for you. So you can use it. Let's say you are reviewing literature. You can use consensus.ai. I can share my screen so that you can see what I'm talking about. Let me share my screen. Let me see which one. Okay. So consensus.ai. This is the consensus.ai. I can put it in the chat box so that you see the app. So let's say you are looking for articles, and then the traditional way you have gone through your library, website, you didn't get much information, you can use consensus.ai to give you some of the articles based on your topic. And then when you get those articles, you can also use maybe ChatGPT if you have GPT Plus or Cloud AI to get ideas in terms of if you want a system to summarize that information for you, you can just click on, you can upload up to five articles if you are using a free version, and a system will give you a summary. The summary will just give you ideas about the article just to see whether you have to read further. So after choosing all your articles, you can also write about them, and then you can use Pipapao, right? This is the software that I use to edit, right? The good thing about this one is that it looks at your paper from the scholarly point of view and edits it for you, right? And it gives you suggestions. It doesn't write things for you. It tells you that, okay, you have to change this word, you have to change this grammar. It recommends things for you, and then you decide whether you want to accept the recommendation or not. So that's what I do. I don't allow the system to do things for me. I just use the tools to help me to accomplish my goal because I am the researcher. The AI is not a researcher, right? So you are supervising the AI system to help you to accomplish your goal. And I think if you do this, if you follow this process, you are not worried about plagiarism, right? You are not worried about, oh, somebody checking and seeing that, oh, you use AI to analyze, or you use AI to write your document. AI can support the process so that it speeds up the process by not doing things for you. So that's what I would do if I were you. All right.

[00:06:58] Speaker 1: That's a great takeaway. If you are just joining in, we are answering all the questions we can fit in within 30 to 40 minutes. So we have just covered how to use AI models to assist, not to do the work for you, but to help you with editing. And Dr. Adu mentioned PayPal and also Consensus. That was the name of it?

[00:07:25] Speaker 2: Yes, Consensus.

[00:07:27] Speaker 1: Those are two other AI apps, models that you can use to help with the editing and also to help you with your literature review. Not to write your literature review now, but with the search. And I like perplexity.ai for that purpose as well because perplexity, for me, is like Google. It's a search engine, but it gives you sources. It cites where the sources are. So I like that for academic purposes. Okay. Dr. Adu, I see you sharing your screen.

[00:07:59] Speaker 2: Yes. You were talking about perplexity. That was I was trying to share my screen. Okay.

[00:08:04] Speaker 1: All right. So another question coming in. You touched on it, on plagiarism. And this question is, how do we check the plagiarism of an AI-generated research script?

[00:08:21] Speaker 2: Yes. So we have many AI plagiarism checkers. I'm hoping that I mentioned this right. I find it difficult to mention state plagiarism. So there are many, and some are over-promising. Some are doing well. But I don't want to recommend the tools for you because I think that if you pass through the steps that I'm giving you, you don't need plagiarism checkers. The same way a GPT can give you wrong information, the checkers can also wrongfully tell you that maybe your information is written by a human being or your information is written by AI. Right? So we shouldn't over-depend on the checkers because the checkers can also wrongly flag your information. So as a researcher, what you have to do is to not allow AI to write things for you. Just gather ideas. One example is that you gather ideas from maybe a GPT. You can also verify those ideas using Perplexity AI. Perplexity will give you the source, and then you write about them. Right? And then you can use PPAPAL to edit your documents. Then you are not going to worry about using any checker because all these checkers, they are also AI. They can make mistakes. There was a story where a professor accused a group of students that they have used AI to write their documents, and it wasn't true. AI wrongfully gave you wrong information. And then some of the checkers, they are over-promising. Oh, this one will even change everything, write it the same way like a human being, and then you check again. And why are you going through all this stress by spending your time and energy writing things yourself, using AI to help you to write, and then you can edit your document using AI? Right? I think that will give you, you know, it will not give you the stress or worry about, okay, somebody is going to find out whether this document is written by AI. And also you have to be transparent, especially if you use AI to analyze your qualitative data. You can say that, you can say that the same way that you use SPSS to analyze your quantitative data, the same way that you use InVivo to analyze your data, you can say that, I use maybe ChartGPT as a data analysis tool to go through my transcript, identify information that are significant. When I got the themes or the codes, I, as a researcher, review them, making sure that they represent the information that you got. Why don't you do that? And, you know, we're going to talk more about this one when, you know, you join us. We have a masterclass coming up on Monday. You can join us and I'll be able to give you all the information that you need to help you to be successful when you are writing your dissertation or doing your research.

[00:11:40] Speaker 1: Yes, yes. So I know the banner is on the screen, but we have to say that the masterclass is full, Dr. Aduf.

[00:11:50] Speaker 2: Yes.

[00:11:50] Speaker 1: The masterclass this week is full, but we can squeeze in two more spots. But first, you have to be quick because we know the interest is overwhelming. It's full, but we have two more spots we can accommodate. We can squeeze it in if you're quick, so hurry and sign up. And the question, a question in the chat, you did not get the name of the software used to edit AI. It's paperpal. P-A-P-E-R-P-A-L. It's on the screen. Paperpal. Okay. Gabrielle has a question here. Is it advisable to rephrase a text in AI and use it in a paper, but still cite it as needed?

[00:12:37] Speaker 2: Yes, you could do that. But the thing is, okay, the best way, I would always say if I were you, right? Why don't you rephrase the text yourself and then use AI to edit it for you? So the system is not taking words. The system is just helping you to make sure that the grammar is good, that you have written in a scholarly way, right? Why don't you do that, right? So if I were you, I would not allow AI to paraphrase or do anything for me. What I would do is that I'll write it myself and then let AI edit it for me, right? And then the good thing about paperpal is that it will show you where it has recommended, right? It's the same thing as a human being editing it for you, and then you decide, you accept or reject the suggestion, right? So I will suggest that you should do that. And the good thing about paperpal is that it also has a plagiarism. I've mentioned this right today? Yes, plagiarism. Check out, right? So you can upload your document there, and then the system can check for you. It has partnership with Turnitin, right? If you are familiar with Turnitin, it's a system where you can also check whether your document is, you haven't copied or you have wrongly cited something or you copied information without giving the right citation, right? Another thing is that the second option is that, yes, you can let AI maybe paraphrase it for you, but according to APA, American Psychological Association, you can cite the source. So as you can see here, you can cite. So let's say you let AI paraphrase it for you. There's a way that you can cite. I can also send this link to the chat and see whether you will get it. So that you can truly cite it. But another thing that you have to think about is that find out from your institution, what are their policies about using AI? Because everything that you have to do, you have to make sure that you are going in line with the policies. If your institution say that don't use AI, please don't use AI. If your institution say that, oh, you can use it, but you can use it as a supporting tool. Yes, you can do that and be transparent. If somebody asks you, how did you use AI to analyze your data? You should be able to tell what the role that AI play in your study so that you not be worried because we don't want to give, we don't want to give all our right and responsibility to AI to manage things for us. Because when you do that, you lose control, right? You'll be a scholar, but you don't know how to write, do a scholarly writing anymore, because AI can do it for you. You should get it like a basic skill, but AI will be helping you to perfect. I always say that AI is like an icing on the cake, right? The cake is your writing. The icing is like, you know, making your writing nice. Then you kind of, you know, let AI help you, right? So you have to be very calculated in terms of using AI for your study so that you can, you'll be able to explain to people how you use it. And I think you will not be in trouble.

[00:16:20] Speaker 1: Yes, thank you. Well said. I like that you gave the caveat, use what your institution recommend. Don't say, Dr. Adu say, I can cite it this way. And thank you very much, Chris. Chris Zielinski. I hope I'm saying your name correctly. Chris Zielinski, President-elect of the World Association of Medical Editors. He has put in the chat the guidelines for how you use chatbots and AI in relation to scholarly publications. So thank you very much, Chris, for that. Dr. Adu, persons are asking for the link to the masterclass in the chat, if you could put that in, because I guess somebody wants to be quick. There's only two spots that we can squeeze in there.

[00:17:07] Speaker 2: So please be quick. So let me copy and paste it in the chat box.

[00:17:14] Speaker 1: Thank you, Gregory. Hopefully you can grab one of those spots. All right. Next question, because it's about answering questions, Dr. Adu. The question is, what are the advantages of using AI over traditional manual coding methods?

[00:17:34] Speaker 2: Okay. So advantages. I think, you know, one thing that comes to mind is it's reduced the frustration of going through the huge, especially when you have a huge amount of data, right? And you don't, you have limited time. You can let AI quickly review in seconds and then give you ideas about the themes and codes and also extract significant information based on the research question that you have. In terms of, you know, quality, yes, AI may not give you a quality information, but I think it's something that you can use to start with. Right? So I think that the main advantage is, you know, reduce the frustration of analyzing the data yourself based on limited time and resources, and also to reduce the time of you coming up with themes and addressing your research question that you have. But I always recommend that you use AI as a supporting tool. So this means that you can do manual coding, manually analyze your data, but AI can help you by bouncing back ideas from the system. Let's say you are going through the data and you are not sure about the kinds of codes or how to phrase the themes and the codes that you are getting. You can ask AI, okay, these are some of my transcripts and this is my research question. Let me give you a little background on my study. Can you suggest themes or codes addressing this research question for me? Right? Then a system can suggest, and then you can use some of the themes to help you to analyze your data. So you can see it as a supporting, playing a supporting role as you are analyzing the data. You can also even use in vivo, Atlas TI, but you can also go back and use maybe chat GP to get ideas about the themes and the code, and then to help you to really analyze your data. Another thing is that sometimes because the information that you are going through is huge, right? You may overlook some of the important ideas. You can tell AI to extract significant information from the data for you so that you can compare with what you have and then maybe there are some things that you overlook that a system can catch for you. So it can also play as like auditing, trying to make sure that you have captured all the information that you have to capture for you to address your research question. So there's a lot of advantages, but I think another thing is that we may reach a time where you don't have to do your analysis manually. You don't have to use maybe some of the qualitative software that are existing. You can focus on using AI, upload your data, ask a system question, you get your results, and then you present your findings. We will reach that stage, but we haven't reached that yet because you can do it, but in terms of credibility, right, will people agree with the findings that you have gotten from AI? For this, people might question your findings, but I think to reach a stage, people will be, AI will be acceptable in terms of using it to analyze, fully analyze your qualitative data.

[00:21:15] Speaker 1: All right. Thank you, Dr. Addo. If you're just joining us, we're answering your questions as much as we can within 30 to 40 minutes. So Gabrielle is up next. Gabrielle asks, do you have any detailed lessons on doing content analysis using AI, just like you have for doing thematic analysis?

[00:21:38] Speaker 2: I don't really have. I haven't done a video I should do. When you go to my YouTube, I have a lot of videos on how to use AI to analyze your qualitative data. But I think it's the same process, right? The content analysis, you already have codes or themes, right? And then you can say that, okay, AI, these are my transcripts, right? These are my codes that I've already generated. Can you go to my transcript, extract information to connect to these codes? You'll be able to do that. It's possible, right? The system can go, especially if you're using GPT-Plus or Cloud AI, you'll be able to, the system will be able to extract significant information and connect them to the predefined codes that you have, that you'll be able to do that. And another thing is that you can, to make it more accurate, you can even define the codes so that the system will know the meaning of the code before you ask the system to extract information from the data to connect to the code, right? It's the same thing as, you know, working with your colleague, you know, you have to define the code so that they know what the codes represent before you can use them to analyze your data. So it's possible you can do that.

[00:22:57] Speaker 1: All right, great. I have a question, if I'm permitted to ask a question.

[00:23:01] Speaker 2: Yes, you can.

[00:23:02] Speaker 1: How do the AI models differ when it comes to in vivo? What's the difference? Why would I go for ChatGPT over in vivo?

[00:23:12] Speaker 2: Okay, so in vivo is a very good tool because you can upload your transcript into the system. You can automatically code your data. They have a system where the system can, you know, extract a theme automatically for you. But the problem is that it might not have anything to do with your research question because the system doesn't take into consideration your research question, right? But if you upload all the transcript into, let's say, ChatGPT or Cloud AI, and the system will be able to extract themes based on the research question, if you give the system a research question and give it background information. So I feel like you can use the two tools together, but I think that if you want to do manual coding using in vivo, so this means that you go to your transcript manually. In vivo, you'll be able to develop themes, and then you'll be able to develop some visual representation that the AI cannot provide to you, right? So there are some things that you can create in vivo that AI cannot provide to you. But in terms of if I have to choose one of them, if I want my findings to be acceptable by the research community, I will personally choose in vivo because everybody knows in vivo, right? But if I want to also leverage AI, I will also, I'm not going to use AI alone. I can ask the system, okay, can you generate themes for me? And I can use the themes to go to in vivo and then look for significant information. So you can use both together for people to accept your information. But I think that, as I said, it will reach the time as people become comfortable using AI, then you can fully use AI to do your analysis. But this time, I feel like using both, right? A little bit of in vivo, maybe a little bit of AI coming together for you to have a very good result.

[00:25:34] Speaker 1: All right. On to the next question because we have to be quick. And the next question is, do you have any suggestions for transcription tools powered by AI?

[00:25:50] Speaker 2: There are many. There are a lot. But one of them that I can recommend is, we call it Rev AI. Let me see whether I can share my screen. Let me see. Let me go here. Transcription. Yeah, so can you see? It's rev.com, right? You can upload. So when you go here, you can upload your transcript or your audio. No, you upload your audio or video, and the system can transcribe it for you, right? And it's not going to be 100%. They said 99, but it's showing that it's not going to be 100%. So you have the chance to also listen to the audio or the video, watch the video, and then make the necessary correction, right? Another one could be Descript. Let me see what I can find in Descript. There are a lot. I always say that just try them out and then see which one is best for you, and then you use it. Descript. Let me see. Yes. So one thing about Descript is that you can edit your video. You can also edit your audio, too, right? It's automatically transcribed for you and can listen to it and do the necessary edits, right? I use this one for my videos if I want to get a transcript, and I can also cut some of my videos and put them on YouTube. So this one is also helpful. So there are many, but these are the two that I personally use.

[00:27:33] Speaker 1: Okay. All right. Indeed, there are many. My personal favorite so far is assembly.ai, and when I have my interviews recorded, I just upload it there, and it gives me the transcript of what was said, and it can do many other things. So it's assembly.ai.

[00:27:54] Speaker 2: Okay.

[00:27:55] Speaker 1: Yes. So once again, we have a master class next week. There are just, well, two spots. We have expanded it because we were full, but we said, okay, the interest is overwhelming, so you have to be quick. Full is full. This time, full is really full. We are not going to accommodate anybody else after this. Just two spots left. On to Santiago. I think Santiago might have the last question. So if you have another question, please put it in quickly as well. So Santiago asks, do you have any recommendations on how to upload the documents, such as transcripts, regarding the form of the documents? Is it better to upload separate documents or one unified document? Are there limits on this in GPT and Cloud? So my interpretation of the question is, if you have 10 documents, do you do it single, or you consolidate and then upload it as one document? What is best?

[00:28:59] Speaker 2: I think that it depends on the limit. So for GPT, I've personally uploaded about maybe 10 documents and then asked a system question, if you are using GPT+. If you are using free version, you cannot upload any documents. For Cloud AI, you can upload up to five documents for free version. I haven't used the paid version, so I don't know how much, but I think you'll be able to upload a lot. If you have a huge amount of data, then think about Google AI, the AI tool called Gemini. You can upload so much. If you have even a book, maybe you have about 500 pages or 700 pages, you can upload and then ask the system questions. So depending on how big you are, we call it contest window, right? How big your data is, you have to choose the one that is really helpful. But for qualitative analysis, simple qualitative analysis, you can maybe upload a few of them and see how the system will provide you some information. And then if you are okay with it, you can upload further based on the limit. But me personally, if it's allowing me to upload all of them, yes, I upload all of them and then ask the system questions so that I'll be able to get what I want. But make sure that you protect participant information. Make sure that you don't use your company's sensitive information and upload them into the system, right? Because sometimes you don't know what the companies are going to use your document for, right? So you have to be very careful. If you are using participant information, try to take out any identifiable information that will cause people to know who your participants are. You always have to protect your participants as you use AI to help you to analyze your data.

[00:31:07] Speaker 1: Well said. Speaking of taking off participant's name and so forth, is your data safe when you use these AI models?

[00:31:17] Speaker 2: Yeah, it depends on the type of AI model that you are using. You have to be very careful. That's why I choose the one I told you earlier about ChargeGPT. ChargeGPT, they are open. That's why their company is called OpenAI. They tell you what they're going to use your data for, right? And how to control your data. So let's go quickly to ChargeGPT, and I'll show you if you are using a paid version, right? So let's go there quickly. And then I will see. Let me see. So when you go to, I use the paid version. When I go here, I can go to settings. And then you have, you can, when you go to data control, you can tell the system that it shouldn't save all your conversations, right? And it shouldn't use any of your data to train your model, right? So I can just turn this one off. If I turn this one off, it will indicate that they will not save your conversation that you have with the system, and they will not use your data to train their system, right? So, you know, for ChargeGPT, you have control over your data, right? But I always think about this one. You always have to think about the risk and the benefit, right? The more they use your data to train the system, the better the model for you as you use it. So if you feel like the data is not all that sensitive, you can let them use it to train so that they have better models for you to use. But if you feel like your data is sensitive and you don't want to get into the wrong hands, you have an option to turn it off. So I like data control when it comes to this one, right? For Cloud AI, they indicate that they are not going to use your data to train their system, right? So they are upfront with you by concerning that. And also, you can always delete the conversation you have with the system, right? So when you go to Cloud AI and have a lot of conversation with the AI, I can go here and say delete, and I'll be able to delete my conversation, right? So you have control. As I said, risk and benefit. You have to think about the risk. When you delete that information and then you come back and you want to have further conversation, the system cannot remember what you have deleted. The reason why sometimes you save this information there is that let's say you want to come back and continue the conversation with the system, although maybe you have uploaded your transcript in the system and then you have interaction and then tomorrow or the next day, you want to come back and have conversation, the system will remember what you have discussed so that it will give you rich information. But if you have deleted your data or your conversation, then you are not going to be able to have – you're going to have fresh conversation with the system, right? So you always have to think about, okay, what's the risk of me saving my information into the system and also the benefits you have to assess. And I think you have to be cautious because, as I said, there's a lot of AI tools that are coming. Some may be doing that to just get your information. So you have to be very careful because now information is the new currency, right? The AI tools are now – they need data to train their system and they have reached a stage where they are not getting a lot of data. So they are moving into synthetic data. They are allowing AI to synthetically generate their own data so that they can train on. Human data are becoming scarce now. So they always want to find ways to gather your information so that they can train their system. So as I always say, you always have to balance. Think about the tool that you are using. Look at how they're going to use your data, right? Always look at the privacy issues and policies so that you'll be able to use it in a way that will benefit you, not only benefit the AI organization.

[00:36:00] Speaker 1: Great. Dr. Adu, we have time for one more question.

[00:36:05] Speaker 2: Okay.

[00:36:05] Speaker 1: But you get to pick of the questions in the comments that you would like to address as last because we're right up on time.

[00:36:14] Speaker 2: You can pick anyone for me. I'll be happy to. I want to make sure that I address all the questions. Let me pick one. Somebody said, is REV, you know, that transcription tool, REV, is it free? It's not free, but you can use it for a few audio, right? And for you to see whether it's good for you before you start paying. So, you know, it has a way of using it a little bit for you to see the outcome before you can upload all your audio for the system to transcribe for you, right? But have we answered all your questions? I want to make sure that we answer.

[00:36:56] Speaker 1: We can squeeze in because we're kind people. So let's squeeze in two.

[00:36:59] Speaker 2: Squeeze in like one or two.

[00:37:01] Speaker 1: Two more into the masterclass if you're just joining in. I mean, time went quickly. Next week, it's a week-long masterclass. We go in depth to show you how to write prompts correctly, how to create an AI assistant, how to use AI in an ethical and responsible way. We discuss data security more in depth. It's going to be hands-on. It's going to be interactive. And we have more time for your questions, but we're kind. So we're squeezing in two more. One is, is it possible to have a codebook or predefined themes before examining the transcripts and then try to find excerpts in the transcripts that can be categorized into these themes?

[00:37:48] Speaker 2: Yes, you can do that. What you are saying is that you call it content analysis, right? We call it coding frame or code frame. You first develop the codes or the themes before you go to the data and then extract information that are significant and then drop it or connect them into the theme that you have already developed, right? We normally do that if you want to use maybe your conceptual framework to go through the data, right? Let's say you have some components and a conceptual framework and you want to see whether they are consistent with the data that you have. You can do that. Let's say you also, you can use literature to the topics and the literature and you use it. But I think the most important thing is that make sure that whatever you are doing, whether you are developing themes before going through the data or developing themes during the data analysis, make sure that they are addressing the purpose of the study or the research question that you have. That's the most important thing. Yes, Gabrielle, you can do that. You can have your predefined codes or themes, go through the data, and sometimes you may develop more themes and then all codes and then categorize the codes to develop your themes. You'll be able to do that. It's possible.

[00:39:06] Speaker 1: All right. Last question now from Maureen. Thanks, Maureen. You have been so active in the chat. The question is, how do you think through qualitative analysis with 30 interviews, where you start with a list of questions, but then iteratively tailor the questions to answer questions more deeply, following the path to get deeper? How do you think through that qualitative analysis? Is that a question you can answer?

[00:39:36] Speaker 2: One minute to ask the last question. Yeah, I think I will try. The thing is that you are going through the data inductively. You are trying to allow the data to drive the kind of questions you're going to answer and also the information, significant information that you extract. It's possible for you to do that. You are going through the data in an open-ended way, allowing it to drive you. But I think that it's going to be too much work for you to do that. Why don't you have specific questions? Let's say two or three research questions and then go through the data, identify information that are significant and use it to address the research questions that you have. And also anything that is not addressing the three research questions that you have, you can also document them as other observation or other findings that you can come back and look at it. But I think it's possible. But the way that you describe it, it's going to take a long time for you to finish the analysis. If you want to do that, I think you have to think about the benefit of doing that. If you want to explore and really address a series of research questions, you can do that. But I think the best way that I've seen is that have about maybe three research questions and go through and then identify information to address them. And then also have anything that is not addressing the research question, kind of put it under other findings or other observation and then look at them later. So that's what I would do if I were you.

[00:41:14] Speaker 1: All right. Thanks, Dr. Odu. We managed to squeeze two in. Thanks, everyone, for your time. Thanks for joining us today. Hope to see you at the master class a week long where we can go in more in-depth information. We'll be covering ChatGPT, Perplexity.ai, Cloud.ai, showing you how to write prompts, how to get better results, and how to do it in an ethically responsible way. Two spots I need to check. Maybe they're already up. Who knows? But go and grab one because we are not making any more allowances for anyone to come and say, please, please sign up now. And that's it. Thank you for your time. Dr. Odu, thank you so very much. And yeah.

[00:41:57] Speaker 2: You're welcome. Thank you for your time, too. I really appreciate it. Thank you, everyone. Bye-bye.