How AI Is Changing Qualitative Research in 2025 (Full Transcript)

[00:00:02] Speaker 1: Hello. So, today, we are speaking of using artificial intelligence for qualitative research. My name is Anmoye Brown, and I'm joined by Dr. Philip Adu, who is an expert on the matter. So, to kick us off, Dr. Adu, like, you have been speaking about the use of AI for qualitative research for quite some time, for more than a year, I believe, right? So, what are some of the changes that you have seen in terms of the software in the last year and a half?

[00:00:44] Speaker 2: Yeah, I think there has been a dramatic change in terms of performance, right? Long ago, like a year or two, when you ask a system a question, you may get general information. But now, I think if you give a very good context, you get a rich information from the system, especially when you are analyzing your qualitative data. And it has also improved in terms of numbers. Now, you'll be able to put your data, quantitative data, into the system and access the system to run the analysis for you and then get the result. And it's becoming more accurate in terms of numbers. Especially, Cloud AI, they are performing very well. I think, as time goes on, they continue to improve. And also, researchers and practitioners are now feeling comfortable using some of the tools and trying to think about how some of them can help them to, you know, complete a task. And I think that we as, you know, researchers have to, you know, explore and see which one will be best for us and then we use it to improve our work.

[00:01:56] Speaker 1: Yes, yes. So, if you are just joining us, we are speaking about how AI, the software, have changed in the last year and a half. So, we would like to know, hear from you a bit. So, if you could just put in the chat, whichever platform you're on, where you're based. And also, what changes have you noticed in the AI models? Right? As Dr. Adu mentioned, the outputs are becoming less generic. And I remember I used to tell persons, do not use ChatGPT for numbers. ChatGPT is no good at numbers. But surprisingly now, you can use ChatGPT for even quantitative data analysis. And a few months ago, it was ChatGPT 3. Now we have ChatGPT 4.0. Was it available?

[00:02:55] Speaker 2: It's so amazing how, you know, there's, you know, there's so much improvement in terms of how, you know, the software is helping us to complete our tasks. And I think, you know, there's a concept called Retriever Augmented Generation. We call it RAG, Retriever Augmented Generation. So, what is happening is that normally AI tools, you know, is a language model. They are good in language, but they are not generally good in numbers. So, what they are doing, the system they have done is that they have connected some programming tools to that, like Python. That will be good in coding and also doing numbers. So, whenever you ask the system, maybe ChatGPT, can you run this analysis for me? Maybe develop create mean or develop standard deviation or what is the correlation between these two variables? The system communicates with a programming tool like Python, run the code, and get all the answers. And then it provides you that information. So, we see how we can call it the program is outsourcing, right? If it cannot perform, it's, you know, connect to other tools so that you'll be able to get what you want. And that's where we are rich concerning numbers and being able to improve in terms of statistics and also doing quantitative analysis.

[00:04:17] Speaker 1: MARIJANA NIELSEN-SHLIZERMAN Quite a leap. So, I see here in the chat, we're joined by Mustafa and also by Vasa, based in Finland. Welcome. And I see Malahat here from D.C. Thank you for being with us. And from Telda. I hope I pronounce your name correctly. Telda from France, right? How is it going there with the Olympics and everything, right? So, Telda says, if it wasn't for Otter, Otter.ai, I couldn't do qualitative. I'm a genius.

[00:04:50] Speaker 2: MALAHAT OTTER Yeah, and I think you hear this information all this post all view all the time. You know, the AI tool has really empowered us. For me, for example, I'm very good in qualitative analysis and qualitative research, but not all that good in quantitative. But because of these tools, now I can perform complex quantitative analysis. So, it's really empowering me as if like I'm now an expert, right? So, you see how powerful these tools, if you use it well, it can help you to accomplish a task and you feel like, wow, I don't have to consult an expert, right? I can just have interaction with the system and get what I want, especially qualitative analysis. You know, you can just put your data in and ask a system question, develop calls, extract significant information to address my research question. Can you develop a table? Can you develop a web cloud? The system can do that for you. So, it's like, wow, how empowered we are in this AI world if you use it all the time and use it in your workplace.

[00:06:03] Speaker 1: Well, Dr. Adu, I'm going to be devil's advocate because, you know, quantitative data analysis wasn't your strength, mine neither, right? But with the use of AI, just like Telga, who is now a genius with it, I'm being devil's advocate. Do we need people with doctorates anymore and PhD? So, persons like you now, all your years of education, getting a doctorate, I mean, with AI, anyone can do it? Is a doctorate still relevant?

[00:06:39] Speaker 2: Yeah, I mean, I think you always need some kind of foundational knowledge, right? It is the same thing as, you know, using AI to do your statistical analysis, right? You should have basic understanding of statistics. How will you know whether the analysis the system is providing to you is the right analysis, right? So, getting basic understanding is so, so important, right? So, that's why we always say AI is not going to replace any human beings, but it's going to also enhance our work. So, PhD, yes, it's needed because, you know, you need to get a skill in identifying a problem, looking into the literature and trying to develop a gap and develop research questions, collect your data, analyze and present your findings in a meaningful way. All these steps are needed. But imagine you have AI on top of that, looking for specific literature for your study. You don't have to spend so much time. With AI, you can easily get relevant articles and then do the analysis and then come up with gaps and then do your research. So, I think that it's making our work very easy. It reduces the stress and also we are becoming more efficient in what we do. So, yes, PhDs is very important.

[00:08:02] Speaker 1: That's the takeaway. The doctorate and the PhD is still of value. Okay. So, welcome Fatiha from Columbus and Ohio and also Kingsley from Ghana. And Malahat had a question, right? If you could comment on language, can you comment on language for document translations? Would it be acceptable for doctoral research when working with diverse communities?

[00:08:36] Speaker 2: Yeah, in terms of language translation, I think that this is where the custom AI tools are very important. So, one example is that using chatGPT to try to translate, you may not get very rich information because it gives you the general translation. It doesn't take into consideration diversity and culture. So, one thing that you have to do is that, is there any tool that has been trained with these kinds of information so that you can use? So, I think there's more about identifying a specific AI tool that has been trained concerning this kind of translation and then you can use it. And there are a lot of translation tools, AI tools that you can explore and see. If you want to use chatGPT, I think what you have to do is to be very, very specific about a context, right? Where is this information coming from? Which country? What is their culture like? So that the system will understand before it will do the translation for you. So, I think looking for a customized one will be the best way, yeah.

[00:09:47] Speaker 1: Okay, all right, great. I see some familiar faces in the chat. Al Habib in Nairobi, welcome. Thank you for joining us. I follow your content a lot on LinkedIn. So, thank you for being here today. And I see Paco Sanchez from Spain. So, thank you for joining. And also, Dr. Adu, I see persons are agreeing with you that a PhD, there's still a need for it, right? So, the consensus is that, yes, we need a PhD still. It's not going to replace academic credentials. Okay, Alberto in Italy is also joining us. Okay, so, we have spoken about how the AI models have advanced or changed over the last year. We speak about new models coming out and also cloud.ai being where I am. I couldn't access it months ago. But has there also been changes in how we are using the users of AI versus how persons are using AI now, researchers, versus a year ago, in your opinion?

[00:10:59] Speaker 2: Yes, you know, I think now, you know, researchers and practitioners are now becoming feeling comfortable. Those who are not even comfortable trying to get ideas about, okay, what is in for me, right? And asking questions. And I think, you know, it has sparked a lot of conversation concerning, which aspect of my research should I use AI, right? Am I going to, how do I cite the source? Should I tell people that I use AI to do my analysis, right? So, these are the questions that people are asking. And also, in terms of software, a couple of AI, a couple of qualitative analysis tools, like InVivo, Atlassia, MassQDA, all of them are incorporating AI into their software so that, you know, you have an option to use it. You have, you know, it's not like the AI would do all of the things for you, but you have an option to develop. So, I think that a lot of people are now feeling comfortable, and some people also are sceptical, and it's always good to be sceptical, because this is what you are helping us to do. You are helping us to make sure that we are using the tool in a way that is proper, right? So, continuously questioning the quality, the efficiency of using the AI tools is very important. It calls us to rethink and try to use in a way that is ethical and also transparent. And I think that in terms of data analysis, especially qualitative data analysis, if you see the tool as a data analysis tool, and then you show how you interact with the system, being transparent, and then you site source the source, I think you will be fine. So it's all about being transparent, and also showing that you are leading the analysis, not a system doing things for you. You have to question, right, what mechanism did you put in place to make sure that the findings or the theme that you have developed reflect what participants are saying? Did you ask the system to reflect, did you ask the system to evaluate the cause and the theme? Did you ask the system to extract significant information to connect to the themes? These are the things that you always have to think about. You always have to be sceptical about the information that you are receiving from AI, too, so that you will be able to get the rich information from the system.

[00:13:38] Speaker 1: MODERATOR Very interesting. But practically, how do you cite when you have used AI? I know contexts differ, and institutions have different rules and regulations, but what is a good practice you would advise if we want to cite from an AI source? DR.

[00:13:57] Speaker 2: OYOLA Yeah, so I'm trying to follow the APA recommendation, right, so when you, one example is when you use it as a data analysis, I think it makes sense when you develop themes from the system, you can cite, right, you cite by indicating the software, and also the year of the software. Sometimes, you can also indicate the model, like is it GPT-4 or GPT-3 that you use? And also, you can also put the conversation, if you have space, you put the conversation in the appendix so that people will see the kind of question you ask the system and the response that you got. In terms of generating ideas, you don't have to cite because it's the same thing as talking to an expert. You may have an expert and have a conversation and gives you ideas, right? But if you are copying and pasting, then you have to do a quotation mark there and then cite the source. So, it also differs from department or institution by institution, so you have to check your institution and see the requirement, and I think because AI is new, we are all new in terms of what is the proper or what is the best practice, so we are learning from each other, so you can also, any audience, any of my audience can also give ideas about what they are doing to cite the source, right? But I think, at the end of the day, being honest and being transparent, I think it's the best way to go, right? I don't like the idea of, oh, how can I use the system so that I'm not going to be caught, right? What about how can I use the system in a transparent way so that it will improve the work that I'm doing, right? I think that is the conversation that we really have to have, right?

[00:15:52] Speaker 1: So, in other words, I can't copy and paste. Is it wrong?

[00:15:59] Speaker 2: It depends on what you're doing, right? If you're doing qualitative analysis and the system has developed themes for you, yes, you can copy the themes and paste, but you're going to indicate that I got these themes from GPT or Cloud AI, right, and also you can provide a conversation. But if you are writing, maybe you're doing a literature review, and then you put your articles in the system, the system went through and generates a review for you, you shouldn't copy and paste, right? You can learn and read about what the information and the writings in your own words instead of copying and pasting. I think that would not be the best practice.

[00:16:44] Speaker 1: Okay. And I got to take away as well that, like, the prompts, the prompts are questions you ask the AI if you can have that in the annex, so it's clear what prompts you use to get the out. Okay. Fair enough. Now, moving on now to AI tools. What, in your opinion, are the best AI tools right now for qualitative data analysis or qualitative research? Yeah.

[00:17:15] Speaker 2: For data analysis, I think that, you know, GPT 4.0 is doing a very fantastic job. They are not all that perfect, but I think that it depends on it all depends on the question that you ask the system, right? Imagine that you have met an expert. You cannot just say, okay, expert, this is my problem, solve it for me. You have to provide expert background information, provide a system, good context. What is your study about? How was the data collected? How many participants do you have? Any background information will help the system to customize the response, right? So, sometimes the AI is good, but if the question that you are asking the system is not clear, it's not all that direct, then you can get maybe general responses. So, charge GPT is very good. Cloud AI is also very good for researchers, right? You can explore that. Apart from that, for qualitative analysis, the tools like Atlas AI, they have a function within Atlas AI called intentional AI coding. I really like this function because like charge GPT, you put in the purpose of your study, right? And a system generates research questions for you. Or you can even suggest a research question that you want a system to use. Then based on the research question, the system goes through your transcript, identify information that are significant, extract themes or codes for you, and put them under the respective research question, right? And then you get your results. And you have the ability to also review and take some of the codes out or change some of the wording of the code. So, that kind of working is like co-piloting, working together with the system to get to your destination is very good. MassQDA is also doing that. Envivo is about to introduce AI into their system, I think, to come this month. So, in terms of data analysis, that's very good. For charge GPT can also be used to generate interview questions, right? Let's say you are not sure about a question that you want to ask participants. You can ask the system, I'm doing this study. This is my research question. This is my maybe conceptual framework. Can you suggest some of the questions that I should ask participants? And this is the profile of some of the participants. And then you can get some examples. So, it's in many ways you can use, depending on the task that you want to complete, you can use AI2. I always say that just explore and see which one will help you to accomplish your task and use it. But you have to be very careful about the AI2s that you have to use in terms of privacy and other things that you have to think about.

[00:20:03] Speaker 1: Okay. Which we're coming to later on in the discussion. And I recommend testing out these different AI models that Dr. Adu mentioned. Personally, I use charge GPT and Cloud.AI. And I find that sometimes the charge GPT handles the prompt better than Cloud.AI. So, I use at least two, right? So, at this point, I think it would be nice, Dr. Adu, maybe to invite persons into the discussion. I see the chat has been active. So, if you have a question for Dr. Adu, you can join the conversation. We'll be sharing a QR code shortly. And it's also in the chat. So, if you are screen ready, wherever you are, you will come up on our screen as well to verbally make a comment or to ask a question. So, I see yes, Ed. Ed is asking, will the recording be available later? Depending on where you're streaming from, there will be a recording either on LinkedIn, YouTube, or Facebook when this event ends. So, welcome to persons. If you're just joining us, we are discussing the role of artificial intelligence for qualitative research and qualitative data analysis. There is a QR code for anyone that wants to come on stage with us, ask a question, or to make a comment, you're more than welcome. Scan the QR code, or it's in the chat. You can click the link and join us. All right. So, here's some questions from Tony. Tony asks, for coding interviews and data analysis, what is your recommendation for AI tools besides Envivo?

[00:22:05] Speaker 2: I think, you know, as I said, you can use chat GPT to do the data analysis. You can also use Cloud AI. If I have to choose between chat GPT and Cloud AI for qualitative analysis, I would choose Cloud. But if you don't have access to that, we can use chat GPT because it's, you know, it's very also good. Right? It all depends on the questions that you ask the system. So, so, when you go to my YouTube channel, I have a lot of videos on how to use AI to do the data analysis. You can try that and see which one will be useful for you. And I can quickly share if I have let me see what I can share the second screen. So, this is my YouTube channel. So, when you go to videos, there's a lot of videos on I think if you go to playlist, I have a lot on chat GPT and plus research. You can click on it and watch any of the videos that is there. That will be useful for you. So, yes. You can use chat GPT. You can use Cloud AI. There are other tools. As I said, you can also use Atlas TI. It has incorporated open AI model GPT 4.0 into the system. So, you can as if you are using chat GPT, you can still use chat GPT within Atlas TI. So, there are a lot of options for you. Sometimes you just have to explore and see which one will be the best. But it all depends on the prompt, the kind of question that you ask the system. And also, document the process because people are going to ask you, how did you arrive at the themes? You should show the process, how you did it. And the themes, you also have to make sure that they are representing the significant information and addressing the research question. You have to review. So, use the AI as like an assistant and you are supervising the assistant. The information that it brings, you have to review and make sure that everything is fine. Sometimes you can use two AI tools and see the results and compare and see which one will be the best. So, it's all about exploring and looking at it. But I think that Cloud AI or chat GPT will be the best way. Go ahead.

[00:24:35] Speaker 1: Yes, go ahead and share.

[00:24:38] Speaker 2: Yes, I want to share something about I developed a chat custom GPT and a chat GPT that will help you to come up with interview questions, right? So, if you can, you tell the system that, okay, this is my the purpose of my study, right? Let's copy the purpose on the research question from here. And then you can see that with this tool and the chat GPT called Qualitative Inquiry Guide, the system will be able to generate interview questions for you. So, you can say, can you generate interview questions for me, right? And then you put your purpose on the study and also the research question and see. So, the system will go through and then it will generate 15 questions for you, right? So, you see how it's not going to be perfect. You just have to review and make sure that everything is right. But see how the system can generate interview questions. It's been trained that it should give you open ended questions. It should give you one question at a time. Simple questions that you can ask participants. And if you are not satisfied, you can ask the system to refine the questions or maybe based on a conceptual framework. So, yes, it's possible for you to do that. I'll put a link in the chat box or when you go to it's sometimes difficult. Okay. So, what I will do is that I'll go to when we finish this conversation, I'll put a link in the comment section so that you can get access to this one.

[00:26:30] Speaker 1: Okay. Great. So, I don't know if there's anyone in the queue lined up. I saw a few persons scanning the QR code.

[00:26:40] Speaker 2: Yeah, I've seen four people, but they haven't shown up yet.

[00:26:44] Speaker 1: Okay. All right. Remember, guys, there's still time to join us. Scan the code, click on the link, and come on board. In the meantime, Al Habib has a question, and that is, are there any resources that teaches you how to write proper prompts and the different type of prompts? And please tell us what a prompt is for persons who might not know.

[00:27:09] Speaker 2: Okay. So, the prompts are the questions that you ask the system, right? And I've done a lot of videos on how to develop a very good prompt, right? But the basic principle is that you have to be direct. You have to be clear about what a system has to ask you. I think about CEO. C means providing a context, right? Provide the system background information. So, let's say you want a system to give you ideas about maybe topics or give you ideas about the methodology that you have to use for your study, right? You have to give a little bit of background, right? So, and then E stands for giving examples. One example that I gave here is that let's say you want a system to develop themes for you. You can give examples of the kinds of themes that you want, and based on that, you'll be able to analyze or you'll be able to get rich information for the system. So, looking at some of my videos will be helpful, but you have to be clear. You have to give examples, if any, and you also have to specify the output. What kind of output do you want? Do you want a system to give you in a table form or in a bulletin form or within 120 words? You have to provide that and the system will help you. I think we have someone here. Yeah, hello. Hello.

[00:28:32] Speaker 1: Before you continue, Dr. Adu, could you take off the QR code so that we can see Osei? Is it Osei or Osei?

[00:28:40] Speaker 3: Yes, Osei.

[00:28:41] Speaker 2: Osei. Osei, yes.

[00:28:42] Speaker 1: Osei.

[00:28:43] Speaker 3: Go ahead, Osei. Okay. So, my name is Michael Osei from Ghana. I'm at Western Michigan University. It's an evaluation center, yes. Thank you so much, Doc, for all the knowledge that you are sharing with us. I've been following you for a while, and I'm very proud of you for my countrymen doing so many things for research and also helping humanity. So, as an early scholar, I have taken a lot of inspirations from you. I'm very much interested in AI. So, I will continue to contact you for more inspiration. Thank you so much.

[00:29:28] Speaker 2: Thank you. I really appreciate your support and also following my work. Thank you so much. Let's continue to have a conversation, and thank you for attending. I really appreciate it. Sure. Thank you. Okay.

[00:29:49] Speaker 1: Wow. That's so nice. You're getting love from Ghana. That was so very nice. So, guys, feel free as well. I like that it was good to just give words of support. Even if it's not a question, it's still nice to take the stage to show some appreciation. So, thanks very much for that. So, I have a question from Tony, and that is, if we had multiple interviews, do you recommend coding them individually with AI, or is there a way to do them all at once?

[00:30:24] Speaker 2: Yes, you can do them all at once because what you're doing is that when it comes to qualitative analysis, it's all about data reduction. Technically, you are summarising what participants have given you, right, so that you come up with things that catapult participant experience or their thoughts, right? So, it makes sense to upload all of the documents into the system, right? ChargeGPT, if you have a paid version, you can upload up to ten. Cloud, you can upload more than that, right? So, just, you know, upload all the documents if it allows you, and then you ask the system questions, and the system will provide you. Also, make sure that you have labels for the documents. You can say P1, P2, P3, because when the system is doing quotation, quoting participants, they can indicate where they got the information from. Did they get it from P1 or P10, right? So, always label your document in a very systematic way or clear way so that the system, when the system quotes or extracts information, you know where the source will be.

[00:31:33] Speaker 1: All right. Thank you. And we're getting a lot of positive feedback in the comments section. I see someone says very interesting discussion, Magdalene, yes, in Ottawa, Canada, saying she's been learning a lot, and improving knowledge, and also the appreciation for the information that you have been sharing. Okay. Is there anyone else on the lineup to take the stage?

[00:32:04] Speaker 2: I don't see anyone yet.

[00:32:07] Speaker 1: All right. So, let me go to the next question on my list, then. And this is a technical question, practical question. So, say I have data. Mm-hmm. What steps can I take as a researcher to prepare my qualitative data for effective analysis using AI?

[00:32:28] Speaker 2: Okay. So, first of all, you have to make sure that before even you prepare your data, you have to decide concerning the AI tool that you have to use, and also look at the privacy information and data security, and see whether it's consistent with what you want. Some of the tools, they will use your data to train the model, right? In terms of using your data to train a model, there's a positive side. The more they train, the more the model becomes good, right? So, if you feel comfortable that it can use your data to train, I think that's okay, but you also have to think about participant information, right? Make sure that any identifiable information has been taken out, right? In terms of their names, where they are located, any kind of thing that can cause you to know who you spoke to, you have to take them out and give them maybe P1, P2, P3, so that a system will not have information about your participant. You can also leave the demographic information, but if you think that putting demographic information to the data can cause people to know who you spoke to, then you have to take the demographic information. So, this is all about data cleaning. You have to clean your data, prepare your data before you upload it into the system, and ask the system questions. So, it's all about, and then if you don't feel comfortable AI using your data to train your model, some of the AI tools, they have options for you to use to opt out, right? Example, ChatGPT, if you go to your account, you can tell the system, don't use my data, right, to train the model. There's a way that you can turn it off, right, so that the system will not use your data to train the model, and I always say that if the information is sensitive, then I recommend that you shouldn't use the AI tool, especially the ones that are online, right? We are closely getting to a stage where you'll be able to download the model into your computer, so that you can use it locally. We haven't reached that yet. The reason why we haven't reached that yet is that you need a lot of compute. You need a lot of your memory and your system on your computer should be large to compute, to process the model. The computers that we have right now, the PCs, is not capable of doing that. That's why we have to do it in the cloud. But as time goes on, new computers will come where you'll be able to download the model onto your computer and then run it yourself. Nobody will see the kind of information that you upload, nobody will see the kind of questions. I think we will reach there, but now, if you think the data is too sensitive, then you have to be very careful uploading it to the system.

[00:35:35] Speaker 1: Okay, so all great tips. So I have an assignment for persons listening in. If you are still here and you want us to continue, write in the chat, still here, right? If we should still continue, write still here, or I'll ask one question and we close and go, right? We want to make the best use of your time. So if you're still here, just write still here. Okay. So while I wait to see if persons are still with us, you mentioned some of the You mentioned some of the challenges a bit and like you have to mitigate.

[00:36:15] Speaker 2: Oh, Joseph is here. All right. I don't know whether a person can... Joseph, are you there? Yes, I'm here.

[00:36:29] Speaker 1: Okay.

[00:36:30] Speaker 2: Yes. Thank you for joining us. Yes. Yes.

[00:36:36] Speaker 4: It's very interesting and I would like to follow you so that I should learn a lot, because after a long period of time, I have not been able to join scholars like you to learn from people that have advanced in academia and because of public service here in Nigeria, I have not had an opportunity to further my studies. So I have interest in starting school again. So this will refresh me and rejig me to make me ready for the tasks ahead. Thank you.

[00:37:17] Speaker 2: You're welcome and also thank you for joining us. Yes, you can always follow us on LinkedIn and you can follow Ann and also you can follow me and also YouTube. I have a lot of videos about research and using, especially doing a qualitative research. You have access to a lot of my videos that would be very helpful for you. And also we have an upcoming webinar if you are interested in knowing more about quantitative analysis, how to use AI. We just want to give people basic understanding of how to use AI to analyze their quantitative data. So if you are interested, you can just take the QR code information and then you can register. And yes, we'll be happy to provide you all the information that you need to help you to be successful in doing research.

[00:38:14] Speaker 5: Lovely. I'm happy to see how I can benefit from that.

[00:38:22] Speaker 2: Yes. So thank you so much for joining us.

[00:38:26] Speaker 1: Yes. Thank you, Jose. Dr. Adu, I see a lot of still here's coming in and one person said still here and not leaving until the end. That's good. That's good. I'm happy you guys are all getting value. And the F doctor, your question, please retype it so it comes on the top of the pile so we can see your question. F doctor, whoever F doctor is type your question, please, so we can see it. So I have another question here from Claude. How do you easily code in Envivo a statement in an interview which could apply to multiple research questions? So Envivo, it's slightly off topic, but can you address it here?

[00:39:17] Speaker 2: Can I ask the question again? Is he asking about how do you use Envivo to analyze your qualitative data or AI? Which one?

[00:39:26] Speaker 1: Envivo, not AI. Envivo to code.

[00:39:29] Speaker 2: Oh, okay. So you're asking about how to use Envivo to let's say you have identified a statement, right? That can address research question one and maybe research question two, right? What do you do? You code it for both, right? But you can change the word a little bit, right? Because you always have to think about so this is the process I have to think about, right? When you identify the significant information, ask yourself what is the participant telling me, right? Because without understanding, you may misquote or you may develop a theme that has nothing to do with what the person is telling you. After that, you ask yourself which research question is this information addressing? Oh, it's addressing research question one and two. Okay. So the first research question, what label or what code should I develop to address my research question at the same time representing that significant information? What about a second research question? What code should I develop to address my second research question and also representing this significant information? So it's possible for you to do that. One information can address more than one research question. It all depends on how you frame the theme or the code addressing that question that you have.

[00:40:43] Speaker 1: Brilliant. So we are also answering questions on Envivo, not just AI. Thank you very much, Dr. Adu. I have found F doctor's question. Finally, I found it. It's if you are using a conceptual framework, how do you prompt GPT to align your framework with your research and or interview questions?

[00:41:08] Speaker 2: So this is what I would do, right? Give you a little background information, right? So you can say that, okay, I'm doing this research about burnout among primary healthcare providers. The purpose of my study is to find out the causes and solutions of burnout from the perspective of these people. And I'm using this conceptual framework, right? When developing themes or when developing questions for me, please consider this conceptual framework. And you can even define what the framework is all about. Don't assume that they know the framework if you just give them the system a name, right? If you have given the system a theory that informs your study, try to define the story, the theory. The more questions, the more information that you give to the system, the better the result. Then when you get the questions, you can tell the system to do self reflections. Considering the question that you have suggested, can you review those questions and make sure that they reflect the conceptual framework that I have provided to you? So the system will review and then provide some more suggestions. So yes, it's possible for you to do that. It's all about having a clear communication with the system. I think we have someone here. Let me remove Joseph again. Oh, a different one. And I think the same. Joseph?

[00:42:45] Speaker 1: It's the same Joseph? Oh, okay. Okay. You went already, I think. Well, it's nice to have you.

[00:42:56] Speaker 2: Oh, okay. Okay. So sorry about that.

[00:42:59] Speaker 1: Okay. So Al-Ahib has a question. And that is, please can you explain the ethics of using AI to analysis and claim the authorship of the analysis?

[00:43:14] Speaker 2: So this is what you have to, you know, it's a challenging topic because AI, as I said, is new in terms of using it for research, right? I think first step, you have to find out from your institution what they think, right? Because it's the policies are different based on your institution that you're affiliated with. One thing that I can say is that each institution with. One thing that I can say is that you just have to make sure that you provide a very good citation in helping you to make sure that you are very clear about what exactly the system you took from the system, right? I think that is the thing that I can think about. Maybe, Anne, do you have any suggestion? Because when it comes to ethics, sometimes it's very difficult to explore.

[00:44:06] Speaker 1: Yeah, I think persons should be guided by their institutions as well, and their specific context, right? And just like as with plagiarism, I know AI is not sentient, is not a sentient being, but think of if AI was a human being, would you just copy somebody else's work? Well, some people do. But just as how just taking somebody's work and claiming it as your own is plagiarism, right? What AI produces, take it and pretend as if it's yours. Yeah, then that is plagiarism. But how you cite ethically and responsibly, that should be guided by your context, your institution, your workplace policy, and also what is allowed. Maybe in some organizations, they don't even want you using AI models because of data security and concerns. Okay, so another question coming in. When it comes to AI tools for data analysis, is there a preferred tool? For example, ChatGPT, Gemini, and Elicit. And I think this question, maybe we can answer it in both ways. So let me ask you in this way, Dr. Adu. It's a question, but I'm changing up the question. What's your preferred tool for qualitative data analysis? And what are your preferred AI tools for quantitative?

[00:45:36] Speaker 2: Okay, so let me share my screen. I think that will be helpful for us. Before I answer that, you know, the one that you are talking about citation, I think if you search, right, APA, ChatGPT, you get access to this kind of information. These are the best practices in terms of citing, right? Yeah, so when it comes to tools, AI tools, what you have to think about is that they are producing a specific output to you, right? But the most important thing is that you have to be transparent and cite the source. Citing will solve the problem and tell people about how you got that information. I think, you know, so this information will be very helpful for you. Okay, so can you repeat the question again about the second one that the person wants me to talk about?

[00:46:37] Speaker 1: So they're just asking about the AI tools that you recommend for qualitative and for quantitative.

[00:46:48] Speaker 2: Okay, for qualitative analysis, I will use ChatGPT, which is GPT 4.0. That, you know, that's what I would recommend. If you even have a quantitative data, you can still use ChatGPT and the system will be good for you. Cloud AI is very good for, let me, it's both good for qualitative and quantitative, right? So if you have both qualitative data or quantitative, Cloud AI can also give you. One thing I like about Cloud AI for the quantitative part is that it can give you interactive visual representation where you can click on and can show you visuals. You can make some adjustment to it. It is so, it has really improved in terms of the output that it provides you. So you can try this one. But if your data is purely quantitative, you can use Julius AI. Julius AI is very good for quantitative data. It does well also in quantitative, but quantitative is very good. One thing I like about this software is that it has more than one model. So you can choose that, okay, you want ChatGPT system to run it for you, or you want Gemini, or you want Meta AI, or I think not Meta AI, we have other one, Lama AI, Lama 3 or something too. So it has like about four or five models within the system where you can choose, or the system can choose for you based on a question that you ask. And then it can give you very good output in terms of visual representation, as you can see here, can give you some visual representation that will be very useful for you. So that's what I like about Julius for quantitative. So if you have quantitative data, I will choose Julius, but if you have both qualitative and quantitative, then either Cloud AI or ChatGPT will be helpful for you.

[00:49:03] Speaker 1: Okay, and we'll be discussing, you'll be showing, having a demonstration of how to use julius.ai at the upcoming webinar. If you could put back that link, Dr. Adu, because we have limited time today, but in a few weeks, Dr. Adu will be showing persons how to use julius.ai and ChatGPT and other tools for quantitative data analysis. So we are right up onto time. I see there are still some questions in the chat, but here's what you do. You like whichever platform you're watching from, joining from or re-watching. Like, follow and subscribe. That way you're notified when we have our next free webinar, where we answer some of the questions that we were not able to get to today. There will be a part two and a part three, but you will be notified as soon as that's coming up. I see somebody's going to have a last word. Chassom, can you hear us?

[00:50:19] Speaker 2: I think he's frozen.

[00:50:26] Speaker 1: Okay. So as I was saying, follow, like, subscribe, so you're notified when we have the next free webinar. Chassom, are you with us? Can you hear us?

[00:50:45] Speaker 2: I think I can hear us. Okay.

[00:50:49] Speaker 1: All right. So let's have back the QR code for the quantitative data analysis webinar for persons who are interested in joining. And I think the last question I have for today to close us off, Dr. Odu, is what, in your opinion, is the future of using AI for qualitative research? You know, what is the next step? In a couple of months, what developments do you forecast will take place?

[00:51:20] Speaker 2: Yeah, I think I see that in the future, you know, it will be very easy for everybody to have their own personalized AI tools, and I think that's going to be a big part of what we're doing, personalized AI tools on your computer. So to address the issue of data security, right? So this means that you can easily go online and download it the same way that you had downloaded maybe Atlas TI or in vivo, you can download it on your computer, and then you can be able to interact with the system. Nobody is watching you, what you are going to ask, right? And you can also have the opportunity to train your system, the model to behave the way that you want it to behave, right? It will also, there will be a time where in terms of data collection, AI will be able to, you know, collect data on your behalf. So this means that you, it's like an agent, right? You instruct the agent, AI agent to go to and identify people who qualify to be part of your study and then interview them based on the questions. And the interview will be more of, maybe it can be like your own voice, right? It can clone your voice and do the same thing or use other people's voice to have a conversation. And the conversation is so seamless that you will not even realize that whether it's AI having conversation with you, right? Because, you know, when we are having normal conversation, we have some parcels there, we can use fill out ways, the AI can have parcels and fill out ways so that you will not even realize that you are talking to AI. So that's where I can see things going. And then the same way that now qualitative software like in vivo, Atlas TI is generally accepted, it will reach a stage that many, many people will feel comfortable using AI to analyze their qualitative data. We have, we are not there yet. This is the stage where educational stage where we are, you know, telling people about how to use it and how to use it in an ethical way. So it will reach a stage that people will feel comfortable and then people will accept it using the software. So that's what I see how AI is going to be. I'm thinking about something that maybe might not be ethical in terms of AI. This is a question to you, right? To all of you audience, right? Do you think that it will reach a stage where AI can generate data for you without you going to the particular participant to collect data? One example is that let's say you give the system a profile. The participant should be 18 and above, you should have X number of years, you should have this, you have that, and I have that. And the system generates, we call it synthetic data for you. Will it reach a stage where we don't have to interview people anymore? Because AI will give us the answers already? Will it reach a stage where maybe having conversation to participants is not needed anymore? These are the questions, ethical questions that we have to ask ourselves, right? Because it takes a lot of time and energy and also sometimes participants also may be busy, right? And not want all the information is so sensitive that if you interview participants, you re-traumatize them, right? So instead of re-traumatizing them, can you give the profile to AI and AI generates the data for you and then you analyze and you are going to be transparent. You're going to say that this is a synthetic data and this is the result, right? Maybe in 20 years time, that's how it's going to be. I don't know whether it's ethical or not, but that's what I see because as time goes on, it's challenging the way that we do things and we see things. And I think this is what is going to happen. It's a conversation we can have, right? Maybe later.

[00:55:40] Speaker 1: Yes, we'll definitely pick up this conversation. Al-Ahib had a question that he's paying consultants and they're charging a lot of money and they use AI to generate reports. Are there ways for him to manage the situation? Al-Ahib, look out for my LinkedIn posts that goes live today where I have a LinkedIn post on software that you can use to detect if it's AI content or if it was written by a human. I publish content on it on eight tools you can use to tell if it's AI generated or human generated. So you can look out for that. Now to everyone still here, if you made it to the end, please type in the comments, in the chat section, I made it. We are about to close up, but if you made it to the end, just type I made it. Dr. Adu, thank you so very much.

[00:56:51] Speaker 2: You're welcome. And I think I want to say something last based on the conversation you are having. You see the reason why we have to know more about AI. If we don't know more about it, and somebody use AI, how are you going to detect that the person that you trust, you are giving a work to, the person use five seconds to finish their work and then collect the money, but the person use AI. Imagine you have no knowledge about that. So this one is an example for us to be abreast with AI tools. We are going to lose if we don't learn more about AI. Even if you don't want to use it, just learn about it. So that when somebody use it, not at all, I think you use this AI. When somebody do a post, I say, this one, the worst that they are using, right?

[00:57:48] Speaker 1: It's really AI. Know your enemy, even if you don't trust AI, but know it so you can identify when other people use it so they can't pull a fast one on you. Okay. Great. I see a lot of persons made it to the end. Thank you for your time. Thanks for joining us. Dr. Adu, thank you so very much. I give you the last word.

[00:58:15] Speaker 2: Oh, I think I already said my last word. My last words is also thank you so much. You have been so wonderful in moderating this. I really appreciate your time. So thank you.

[00:58:26] Speaker 1: Thanks you. Thank you too. Thanks everyone.