Latest / Azadeh and Guy on AI / Using AI in Dissertations

Transcript

Using AI in Dissertations · Azadeh and Guy on AI

• 0:02
  SPEAKER_02: And we're on.
• 0:03
  SPEAKER_02: Hi and welcome.
• 0:06
  SPEAKER_02: This is episode two in season two of Azadeh and Guy on AI.
• 0:13
  SPEAKER_02: And today we have a special surprise.
• 0:15
  SPEAKER_02: We have Inusa Malgubri with us.
• 0:19
  SPEAKER_00: Or Dr. Inusa Malgubri.
• 0:21
  Dr.
• 0:22
  Malgubri has recently graduated from the University of
• 0:26
  Nebraska-Lincoln with a
• 0:29
  fascinating dissertation,
• 0:31
  but this is not what we're talking about today.
• 0:33
  SPEAKER_02: So what are we talking about?
• 0:35
  We thought that maybe it's useful for our listeners to be
• 0:41
  engaged with Dr.
• 0:42
  SPEAKER_00: Malgubri's story about using AI in different ways in his academic life,
• 0:49
  SPEAKER_00: in his personal life.
• 0:50
  SPEAKER_00: Yeah.
• 0:51
  SPEAKER_00: We are really excited about your stories.
• 0:54
  SPEAKER_02: So let's start with kind of an introductory question.
• 0:58
  SPEAKER_02: And that is, how do you think about AI?
• 1:00
  SPEAKER_02: What's your attitude towards AI at this point?
• 1:03
  SPEAKER_01: Okay, thank you, Dr. Training and Azadeh.
• 1:08
  SPEAKER_01: And in the podcast, I'm Guy.
• 1:11
  SPEAKER_01: Okay.
• 1:14
  SPEAKER_01: It would be harder for me to do it right now.
• 1:17
  SPEAKER_02: Yeah, but the audience is cheering for you to be able to get it right.
• 1:22
  SPEAKER_01: So AI, I think AI is an excellent solution.
• 1:29
  especially Gen.AI for education and researchers around the
• 1:34
  world because there are
• 1:35
  SPEAKER_01: things that we might have taken three weeks or even months to do.
• 1:42
  SPEAKER_01: And thanks to Gen.AI, we can do it in no time.
• 1:46
  SPEAKER_01: And as you know, AI is as good as you are.
• 1:51
  SPEAKER_01: So I see AI as a factor.
• 1:54
  SPEAKER_01: back to that GAS,
• 1:55
  like I would say an exponent and GAS multiplies the abilities
• 2:02
  and the knowledge
• 2:03
  SPEAKER_01: already have.
• 2:04
  SPEAKER_01: So it's just a tool,
• 2:07
  SPEAKER_01: an awesome tool that we can use as educators,
• 2:12
  SPEAKER_01: as researchers to make our job easier.
• 2:15
  SPEAKER_02: Is there anything about AI that you're worried about?
• 2:22
  SPEAKER_01: The only thing I might be worried about AI is the cognitive process.
• 2:28
  SPEAKER_01: Because when we are doing something,
• 2:31
  what makes it stick in our brain is usually the cognitive
• 2:36
  process that led to the
• 2:39
  SPEAKER_01: answer or the findings that we have.
• 2:42
  SPEAKER_01: And if you use AI throughout the process,
• 2:45
  we might lose that cognitive process that happens in our brain
• 2:49
  that is not visible
• 2:50
  to anyone but that helped make things stick in our brain and
• 2:55
  that make it um
• 2:57
  retrieval easier later in the future in case we have to talk
• 3:01
  about what we have
• 3:02
  SPEAKER_01: done or what we did
• 3:05
  But if you use too much AI and doing research or writing papers
• 3:10
  or producing
• 3:11
  SPEAKER_01: anything,
• 3:12
  you might lose that cognitive ability to retrieve knowledge
• 3:17
  that you produce in the
• 3:18
  SPEAKER_01: past.
• 3:19
  SPEAKER_01: That's the only thing that worries me when it comes to getting AI.
• 3:22
  SPEAKER_02: All right.
• 3:24
  And this is definitely something that has been at the forefront
• 3:28
  of our thinking.
• 3:28
  And that is the unfortunate discussion around what has been
• 3:36
  termed cognitive debt.
• 3:38
  SPEAKER_02: Right.
• 3:39
  SPEAKER_02: Which is a weird way to say this.
• 3:40
  You don't owe that money to anybody,
• 3:43
  that knowledge to anybody except yourself.
• 3:45
  SPEAKER_02: As you pointed out,
• 3:47
  SPEAKER_02: it's like your own retrieval,
• 3:48
  SPEAKER_02: your own responsibility is the core around this.
• 3:54
  SPEAKER_02: But at the same time,
• 3:56
  SPEAKER_02: there is that notion that you can,
• 3:58
  SPEAKER_02: that that cognitive process is almost slipping away,
• 4:02
  SPEAKER_02: right?
• 4:02
  SPEAKER_02: You lose control.
• 4:03
  SPEAKER_02: The more you give AI, the less control you might have.
• 4:08
  SPEAKER_01: Yeah, the less control you might have.
• 4:09
  SPEAKER_01: But it depends on also how you use it.
• 4:12
  SPEAKER_01: For example, I have...
• 4:15
  When I was doing my Google certification,
• 4:18
  I came across an acronym that I love,
• 4:21
  PART.
• 4:23
  When you want to use AI,
• 4:24
  gain AI to do something,
• 4:26
  you need to think of a persona.
• 4:29
  SPEAKER_01: And then the aim, what do you want to achieve with that AI?
• 4:34
  And after the persona and the aim,
• 4:37
  then the recipient,
• 4:38
  who is it intended for?
• 4:42
  SPEAKER_01: And then after that,
• 4:43
  now you think about the tool that you're going to use or the
• 4:46
  theme that you want to
• 4:48
  SPEAKER_01: have it,
• 4:48
  SPEAKER_01: and then the structure that you want.
• 4:51
  SPEAKER_01: But the cognitive death might be...
• 4:55
  SPEAKER_01: let's say decrease,
• 4:56
  that means we can still have some cognitive process using AI if
• 5:02
  we follow this
• 5:04
  SPEAKER_01: process.
• 5:04
  Because to have a persona,
• 5:06
  you actually need to know what that persona can do.
• 5:09
  Second,
• 5:10
  the aim,
• 5:11
  what you want to achieve,
• 5:12
  you need to know where you're going.
• 5:14
  SPEAKER_01: Yeah.
• 5:14
  SPEAKER_01: And you just,
• 5:15
  SPEAKER_01: just like if you're traveling from Lincoln to Omaha,
• 5:18
  SPEAKER_01: you just want to get to Omaha.
• 5:19
  SPEAKER_01: So if you have a car, you go with your car.
• 5:22
  SPEAKER_01: You're not going to walk.
• 5:23
  SPEAKER_03: Yeah.
• 5:24
  SPEAKER_01: And then recipient, where do you want to get the final product?
• 5:29
  SPEAKER_01: It's going to be given to someone or the audience.
• 5:33
  SPEAKER_01: If I research it for the academic community.
• 5:36
  SPEAKER_03: Yeah.
• 5:36
  SPEAKER_01: And then the tone.
• 5:38
  SPEAKER_01: and all the theme that you want to use when you are doing it.
• 5:41
  And you are going to think through,
• 5:43
  how do I want to sound when I deliver this product?
• 5:47
  SPEAKER_01: How do I want it to look like?
• 5:49
  SPEAKER_01: How do I want people to read it?
• 5:51
  SPEAKER_01: How do I want them to say it?
• 5:53
  SPEAKER_01: That's where you put all those.
• 5:55
  SPEAKER_01: There's still some cognitive processes there.
• 5:57
  SPEAKER_01: And then the structure.
• 5:59
  SPEAKER_01: How do you want the GNI to structure your thought?
• 6:03
  SPEAKER_01: And...
• 6:04
  SPEAKER_01: If it produces something that does not resemble who you are,
• 6:11
  SPEAKER_01: of a persona you have in mind when you are giving the prompt,
• 6:15
  SPEAKER_01: then do some tweaking.
• 6:17
  SPEAKER_01: But if you don't have,
• 6:18
  then that's where the cognitive process might not be lost
• 6:22
  completely because you
• 6:24
  SPEAKER_01: are still there,
• 6:25
  SPEAKER_01: you're still at the beginning.
• 6:27
  SPEAKER_01: Yeah.
• 6:28
  SPEAKER_01: In the middle and at the end of a process,
• 6:31
  you think about the thing you want to achieve and how you want
• 6:36
  to sound and how you
• 6:37
  SPEAKER_01: want it to look like and how you want the recipient,
• 6:41
  SPEAKER_01: the audience to feel what you want them to know.
• 6:46
  You think about all those and when you're writing the poem for
• 6:49
  Virginia and then
• 6:52
  SPEAKER_01: the structure that you want it to look like.
• 6:54
  SPEAKER_01: And sometimes you can use your own writing
• 6:58
  SPEAKER_01: If you have previous article, okay, this is how I write my articles.
• 7:03
  SPEAKER_01: This is how I write.
• 7:06
  SPEAKER_01: This is how I go about my literature review or my methodology.
• 7:11
  SPEAKER_01: This is how I do things.
• 7:13
  SPEAKER_01: So then the structure needs to follow these different steps.
• 7:17
  In this case,
• 7:18
  we might lower our cognitive,
• 7:23
  let's say,
• 7:24
  charge that we have and also
• 7:27
  still increase the cognitive process that is going to happen
• 7:32
  during the process and
• 7:33
  SPEAKER_01: lower what we call the cognitive debt.
• 7:35
  SPEAKER_02: Yeah.
• 7:36
  And this is one of the things we've been talking about is
• 7:41
  you've got to make sure,
• 7:43
  especially we're going to go into some questions about the
• 7:45
  research process and the
• 7:47
  SPEAKER_02: academic process,
• 7:49
  SPEAKER_02: keeping the human in the loop.
• 7:50
  SPEAKER_02: We cannot be detached from that loop.
• 7:52
  SPEAKER_02: And basically what you've described is
• 7:54
  SPEAKER_02: There's a person,
• 7:55
  SPEAKER_02: the researcher,
• 7:57
  SPEAKER_02: the reader,
• 7:58
  SPEAKER_02: the writer is there at every step,
• 8:01
  SPEAKER_02: being part of that step.
• 8:03
  SPEAKER_02: You don't just let that step run on its own.
• 8:05
  You define it and then you follow it and then you examine it to
• 8:09
  make sure that
• 8:10
  SPEAKER_02: you've got something.
• 8:11
  SPEAKER_01: As a researcher,
• 8:12
  SPEAKER_01: if you don't follow them,
• 8:15
  SPEAKER_01: your final project,
• 8:17
  SPEAKER_01: the article will be identity-less.
• 8:21
  SPEAKER_02: Well, that, but also would it be yours?
• 8:24
  SPEAKER_02: I mean, in some ways, would it be yours?
• 8:25
  SPEAKER_02: If AI can produce all of that, why do we need the person?
• 8:28
  SPEAKER_01: So it would be identity-less.
• 8:29
  So then it's no longer you because you lose your agency and
• 8:34
  then you apply your
• 8:35
  SPEAKER_01: agency throughout the process.
• 8:37
  Then the final result will still look like what you would do it
• 8:42
  in the best way possible.
• 8:44
  SPEAKER_01: All right.
• 8:46
  SPEAKER_00: Our next question is about the use of AI in your dissertation.
• 8:50
  SPEAKER_00: So could you tell us how you used AI for that?
• 8:54
  SPEAKER_01: I don't know,
• 8:55
  SPEAKER_01: but I know my advisor told me that,
• 8:57
  SPEAKER_01: okay,
• 8:58
  SPEAKER_01: don't use AI to write it,
• 9:00
  SPEAKER_01: but you can use it for polishing.
• 9:03
  SPEAKER_01: Yeah,
• 9:03
  SPEAKER_01: and she said you could use it for polishing,
• 9:07
  SPEAKER_01: which was because I was like,
• 9:08
  SPEAKER_01: okay,
• 9:09
  SPEAKER_01: I have used AI,
• 9:10
  SPEAKER_01: I know how to use it,
• 9:13
  SPEAKER_01: and if you tell me not to use it,
• 9:16
  SPEAKER_01: I might do things and I will not be honest with you.
• 9:19
  SPEAKER_01: All right.
• 9:20
  SPEAKER_01: This is not...
• 9:20
  SPEAKER_01: This is something we...
• 9:21
  SPEAKER_01: And honesty counts in this case because you are doing research.
• 9:24
  SPEAKER_01: We need to be honest in all the processes that we have used.
• 9:30
  SPEAKER_01: And now it's...
• 9:31
  We are now seeing dissertation where people have chapters
• 9:34
  dedicated to how they use
• 9:36
  SPEAKER_01: AI in the writing process.
• 9:38
  SPEAKER_01: But when I was doing it, I basically used it, especially when I was...
• 9:44
  SPEAKER_01: cleaning my data.
• 9:47
  SPEAKER_01: And for example,
• 9:48
  I had some interviews and people spoke French and English and
• 9:54
  everything was mixed.
• 9:56
  SPEAKER_01: And I think,
• 9:58
  SPEAKER_01: I don't remember the transcription software that I paid,
• 10:02
  SPEAKER_01: I first paid to transcribe the interviews.
• 10:05
  And I didn't like how it looked like because I read the
• 10:09
  transcript and I listen.
• 10:11
  and I didn't like it how it was then you mean something was
• 10:16
  missing or something
• 10:18
  was missing so the transcription was not it it wasn't only
• 10:21
  accurate but it's
• 10:24
  It's completely,
• 10:24
  I would say,
• 10:26
  60% of what output I expect,
• 10:30
  which is not what I needed.
• 10:32
  SPEAKER_01: And then I went online and I know my advisor said, okay, don't use it.
• 10:37
  SPEAKER_01: And then I go look for MaxQDA.
• 10:40
  SPEAKER_01: And then MaxQDA 2024 had an AI assistant within it.
• 10:45
  SPEAKER_02: Just a second.
• 10:46
  SPEAKER_02: MaxQDA is not yet a sponsor for this show.
• 10:50
  SPEAKER_02: So MaxQDA, people.
• 10:52
  SPEAKER_01: Okay.
• 10:52
  SPEAKER_01: Thank you.
• 10:53
  SPEAKER_02: We're looking forward.
• 10:54
  SPEAKER_01: They have an AI assistant in it.
• 10:57
  And then I have a UNL license,
• 11:01
  MaxQDA,
• 11:02
  which did not include that AI assistant.
• 11:04
  So I have to go and buy my own MaxQDA where I will have an AI
• 11:11
  assistant that will
• 11:12
  SPEAKER_01: help me transcribe my interview data.
• 11:17
  SPEAKER_01: And then when it's transcribed, I also have it within MaxQDA.
• 11:20
  SPEAKER_02: Yeah.
• 11:21
  SPEAKER_01: And then I have my own
• 11:23
  SPEAKER_01: They now,
• 11:24
  SPEAKER_01: because I conducted interviews myself,
• 11:26
  SPEAKER_01: so pre-workshop interviews,
• 11:29
  SPEAKER_01: post-workshop interviews.
• 11:31
  SPEAKER_01: So I have an idea of what was going on,
• 11:34
  SPEAKER_01: what is the common themes,
• 11:36
  SPEAKER_01: and I look through the transcript.
• 11:38
  SPEAKER_01: And I've done that on a paper,
• 11:42
  and then every day I was coming here in Carlin Pope Edward Hall
• 11:46
  every day from 3 to
• 11:47
  SPEAKER_01: 8.
• 11:47
  SPEAKER_01: Yes, the manual work.
• 11:51
  SPEAKER_01: Manual work.
• 11:52
  SPEAKER_01: And then after that,
• 11:53
  Thank you.
• 11:54
  SPEAKER_01: I wanted to make sure that I got everything right.
• 11:56
  SPEAKER_01: So,
• 11:57
  SPEAKER_01: and I also,
• 11:57
  SPEAKER_01: since I have a transcript within MaxQDA,
• 12:00
  SPEAKER_03: I was like,
• 12:01
  SPEAKER_01: hey,
• 12:01
  SPEAKER_01: I see what might the common themes or quotes be for this research.
• 12:07
  SPEAKER_01: And I took a look and I see how it,
• 12:11
  SPEAKER_01: one of the themes and with different colors,
• 12:15
  SPEAKER_01: which even look nicer than my paper,
• 12:18
  SPEAKER_01: like the themes.
• 12:20
  SPEAKER_01: And I could see that 80% of them were the same.
• 12:25
  SPEAKER_01: But I also had two of my friends that I asked to,
• 12:31
  SPEAKER_01: now Dr.
• 12:31
  SPEAKER_01: Hector Palala and Rafael,
• 12:33
  SPEAKER_01: I asked them to also code some part of my interview data.
• 12:39
  SPEAKER_02: So what you're saying is, so I'm hearing three things.
• 12:43
  SPEAKER_01: AI is now our fourth person.
• 12:45
  SPEAKER_02: A fourth quarter.
• 12:47
  So you did multiple verifications,
• 12:51
  and AI is part of that level of verification.
• 12:53
  SPEAKER_01: Part of that level of verification.
• 12:54
  I did my coding and theming,
• 12:58
  and then I asked two of my colleagues to do it,
• 13:01
  and they did it.
• 13:01
  SPEAKER_01: And then I asked...
• 13:03
  SPEAKER_01: AI to do it and then I have four.
• 13:05
  SPEAKER_01: Yes.
• 13:05
  SPEAKER_01: Then I say, okay, these are the common themes.
• 13:09
  So I use the AI in this case for verification and also for the
• 13:13
  trial because the
• 13:15
  SPEAKER_01: new AI is what enabled masculinity to transcribe interviews and
• 13:20
  SPEAKER_02: And so here's a few things, the few other things that I've heard.
• 13:25
  SPEAKER_02: One is, people, sometimes it's worth paying for the stuff.
• 13:30
  SPEAKER_02: The paid version is considerably better.
• 13:34
  And so if you are considering using AI,
• 13:37
  at least play with the paid version.
• 13:40
  SPEAKER_02: and see if you're getting something better.
• 13:41
  SPEAKER_02: Because 60% accuracy on an interview is absolutely unacceptable.
• 13:46
  SPEAKER_02: I mean, you basically have to redo it.
• 13:47
  SPEAKER_02: It's still better than nothing.
• 13:49
  SPEAKER_02: But it leaves a lot of work to be done.
• 13:51
  SPEAKER_02: When you get to the upper 90s, that's where we want to be.
• 13:55
  And that's where paying for something sometimes is worth it,
• 14:00
  even for a limited time.
• 14:01
  SPEAKER_02: So if you are a graduate student, that's something to consider.
• 14:04
  SPEAKER_02: That's one thing I heard.
• 14:06
  SPEAKER_02: The second thing I heard,
• 14:07
  SPEAKER_02: and that's really important,
• 14:08
  SPEAKER_02: it was important for you to listen and to be very,
• 14:12
  SPEAKER_02: very familiar.
• 14:13
  SPEAKER_02: And one way people used to do it is through the transcription process.
• 14:17
  SPEAKER_02: However, that process is painful and very, very slow.
• 14:21
  And this is one of those things that AI can just take that work
• 14:26
  off your back.
• 14:27
  No,
• 14:27
  I just sit down and watch like a YouTube video because I have
• 14:30
  my transcript.
• 14:31
  SPEAKER_01: I have the audio.
• 14:32
  SPEAKER_01: I have a transcription.
• 14:33
  SPEAKER_01: Then I just plug in my...
• 14:37
  And then I just look at the transcript and listen to what my
• 14:41
  interviewee was saying.
• 14:43
  And then I look at the transcription and then we just correct
• 14:45
  it in real time.
• 14:47
  SPEAKER_02: Yes.
• 14:48
  SPEAKER_02: And that is a powerful experience.
• 14:52
  And it also allows us to deal with more data in reasonable
• 14:56
  amount of time.
• 14:57
  SPEAKER_02: Because if somebody else, something else is transcribing.
• 15:02
  SPEAKER_00: I have another question about your qualitative coding process.
• 15:05
  SPEAKER_00: You said that you did it by yourself.
• 15:07
  SPEAKER_00: Then you asked two other scholars to do the same thing.
• 15:09
  SPEAKER_00: And finally, you asked AI.
• 15:11
  SPEAKER_00: So how much AI was helpful in that process?
• 15:14
  Because you basically did all of the things that we normally do
• 15:17
  when we do qualitative coding.
• 15:19
  SPEAKER_00: Why did you feel that AI's help is needed at the end of the process?
• 15:24
  SPEAKER_01: I wanted my data to be as accurate as possible.
• 15:29
  SPEAKER_01: We are humans.
• 15:32
  SPEAKER_00: You wanted to make sure that you are not missing anything?
• 15:34
  SPEAKER_01: Yeah, I want to make sure that I'm not missing anything that my...
• 15:39
  My study participants said I didn't want to miss on any theme
• 15:44
  or any code.
• 15:45
  SPEAKER_01: That might be important.
• 15:46
  SPEAKER_01: That the AI might be able to catch that we weren't able as humans.
• 15:52
  SPEAKER_01: For example, I know the culture.
• 15:55
  SPEAKER_01: Maybe I might be biased on how I code the data.
• 16:00
  I have a colleague of mine who had no idea what the culture is
• 16:05
  and one who is also
• 16:06
  SPEAKER_01: kind of closer to the culture.
• 16:08
  SPEAKER_01: And now the AI has,
• 16:11
  SPEAKER_01: in Masculia,
• 16:12
  SPEAKER_01: we don't have my persona in it,
• 16:13
  SPEAKER_01: so it doesn't have my persona within Masculia,
• 16:17
  SPEAKER_01: so it doesn't have my background.
• 16:19
  And then it's kind of like blindly coded it without the
• 16:23
  cultural understanding of
• 16:25
  SPEAKER_01: the interview so that I can see how
• 16:30
  The themes are and if my cultural insider has not impacted my
• 16:39
  coding and beaming.
• 16:41
  SPEAKER_01: So that makes sense.
• 16:43
  And how much or how many themes did AI come up with when you
• 16:47
  did that process?
• 16:49
  SPEAKER_01: A, I was able to come up with four themes, more than I had.
• 16:53
  SPEAKER_00: Four themes.
• 16:54
  SPEAKER_01: More than I had and two that I had and that A, I did not.
• 17:00
  SPEAKER_00: Okay.
• 17:00
  SPEAKER_01: So, yeah.
• 17:01
  SPEAKER_00: Yep.
• 17:03
  So the other thing that I heard you say very early on that I
• 17:07
  think is important if
• 17:08
  you're a graduate student and you're thinking about how do I
• 17:11
  use AI in my process?
• 17:13
  SPEAKER_02: Talk to your advisor.
• 17:14
  SPEAKER_02: Make sure you know the boundaries.
• 17:16
  You don't want to find out when you're done that what you've
• 17:19
  done is unacceptable
• 17:20
  SPEAKER_02: to the committee.
• 17:21
  SPEAKER_02: This is a squishy time.
• 17:23
  SPEAKER_02: I think most universities are getting some guidelines around that.
• 17:28
  SPEAKER_02: But it's very,
• 17:29
  SPEAKER_02: very important as a master's students or as doctoral students,
• 17:32
  SPEAKER_02: even as an undergraduate,
• 17:33
  SPEAKER_02: if you're writing a research paper,
• 17:36
  SPEAKER_02: to make sure that the use of AI is understood and that it's okay,
• 17:41
  SPEAKER_02: right?
• 17:42
  SPEAKER_01: You need to always check with your advisor.
• 17:44
  SPEAKER_01: I didn't wait for her to tell me anything.
• 17:46
  I just asked her,
• 17:47
  okay,
• 17:48
  you know,
• 17:48
  I also have a specialist in educational technologies.
• 17:53
  SPEAKER_01: And you know that, okay, I use those technology tools.
• 17:57
  SPEAKER_03: Mm-hmm.
• 17:59
  SPEAKER_01: What are the boundaries?
• 18:00
  SPEAKER_01: Yeah.
• 18:00
  And she told me,
• 18:01
  okay,
• 18:02
  definitely I'm not really fan of,
• 18:04
  great fan of AI in the research.
• 18:07
  SPEAKER_01: Yeah.
• 18:07
  SPEAKER_01: But you can use it for the policing.
• 18:10
  SPEAKER_01: You can use mask QDA, but do not use it to, do not use chat.
• 18:16
  Don't use chat GPT to,
• 18:18
  let's say,
• 18:19
  find the themes,
• 18:20
  code it for you and write everything.
• 18:24
  SPEAKER_03: Yeah.
• 18:25
  And so you need to discover because there are ways to do that
• 18:28
  with reliability.
• 18:29
  I mean,
• 18:29
  we talked in season one,
• 18:31
  we talked about papers that showed how you can do it reliably.
• 18:34
  I love the fact that you did something we did talk about in the
• 18:38
  first season as well.
• 18:39
  SPEAKER_02: And that is that AI can check against biases.
• 18:42
  SPEAKER_00: Yeah, I was going to say that.
• 18:44
  SPEAKER_00: Yeah, we talked about researchers bias in our first season.
• 18:47
  SPEAKER_00: Yeah.
• 18:48
  SPEAKER_02: And so that's a beautiful way to use it.
• 18:50
  So if you had to give a graduate student advice right now about
• 18:55
  using AI in the
• 18:56
  SPEAKER_02: research process,
• 18:58
  SPEAKER_02: what would be kind of the top two,
• 19:00
  SPEAKER_02: three things that you would say,
• 19:01
  SPEAKER_02: here are some things to think about that can be really useful for you?
• 19:06
  For example,
• 19:07
  some grad students usually have a hard time finding the right
• 19:12
  literature.
• 19:14
  SPEAKER_01: And AI can really help you to...
• 19:18
  SPEAKER_01: find some little data that you haven't thought you would find.
• 19:24
  But the caveat is that you also need to be careful because some
• 19:27
  of the AI can give
• 19:29
  SPEAKER_01: you some just fabricated references,
• 19:32
  SPEAKER_01: but you need to make sure that when you tell the AI,
• 19:35
  SPEAKER_01: okay,
• 19:35
  I am doing my research on this topic or my dissertation or my
• 19:42
  master's thesis,
• 19:44
  SPEAKER_01: I would like some research.
• 19:48
  SPEAKER_01: that were published between 2020 to 2026.
• 19:56
  SPEAKER_01: And maybe if you want, you can even use the booleans.
• 20:02
  SPEAKER_01: The title,
• 20:03
  SPEAKER_01: the research title must include those ones,
• 20:05
  SPEAKER_01: or maybe they must have this in the abstract.
• 20:08
  SPEAKER_01: And AI is going to get you most of them.
• 20:13
  SPEAKER_01: And now with links, and then you need to take your time
• 20:17
  Click on each link to actually make sure that what it is said
• 20:23
  is actually what the
• 20:24
  SPEAKER_01: content is about because it happened.
• 20:26
  SPEAKER_01: I already use it.
• 20:27
  And then it happened that sometime it gave you a link that has
• 20:32
  nothing to do with
• 20:34
  SPEAKER_01: what the title is.
• 20:35
  SPEAKER_03: Yeah.
• 20:35
  SPEAKER_01: Yeah.
• 20:37
  SPEAKER_01: It gives you a DOI to another research article,
• 20:41
  SPEAKER_01: and you wouldn't want to look down during your defense.
• 20:45
  SPEAKER_01: Or be wrong.
• 20:46
  SPEAKER_01: Or be wrong, or just not check anything.
• 20:49
  SPEAKER_01: So you can use it for that one.
• 20:52
  SPEAKER_01: Second, you can also use it for brainstorming.
• 20:55
  SPEAKER_01: Sometimes we want something to kickstart our writing.
• 20:57
  So sometimes you know where you want to go,
• 21:00
  but the ideas that you need to...
• 21:05
  SPEAKER_01: So start the framework and the AI can help you give you with a roadmap.
• 21:12
  SPEAKER_01: The AI can give you a roadmap on what you need to do,
• 21:15
  SPEAKER_01: the steps that you need to follow to get there.
• 21:19
  SPEAKER_01: Then it can give you an architecture of a paper that you want to write.
• 21:23
  And then you take the paper that you did with the literature
• 21:27
  and then you have the papers.
• 21:31
  SPEAKER_01: And after that, now you need to then read.
• 21:35
  SPEAKER_01: let's say the abstract,
• 21:38
  SPEAKER_01: but usually if you have done the first verification,
• 21:40
  you didn't need to do that to kind of like declassify the
• 21:43
  article that have nothing
• 21:45
  SPEAKER_01: to do with your research.
• 21:47
  SPEAKER_01: And basically,
• 21:50
  SPEAKER_01: some people say I'm a fast reader,
• 21:52
  SPEAKER_01: but when I take something to read,
• 21:54
  SPEAKER_01: I read the abstract,
• 21:55
  SPEAKER_01: the purpose of a paper,
• 21:58
  SPEAKER_01: skip the literature, go to the methods.
• 22:02
  After the methods,
• 22:04
  I just came through the findings and I go to the discussion.
• 22:09
  And then if I find it interesting enough,
• 22:12
  then I can go back to the whole thing.
• 22:16
  SPEAKER_01: That's how I...
• 22:17
  do things and and with ai now you can even get that one for
• 22:22
  example with uh i don't
• 22:24
  know with adobe reader it can give you a clear idea of what you
• 22:29
  have in that and
• 22:30
  even if you have your papers in google drive if you have your
• 22:34
  articles in the
• 22:35
  google drive when you open the document it can give you a short
• 22:40
  summary of what it
• 22:42
  SPEAKER_01: is and then you will see if okay is it worth reading then
• 22:46
  SPEAKER_01: you read, if it's not worth, you just skip and continue.
• 22:49
  SPEAKER_01: So that's the second thing.
• 22:50
  SPEAKER_01: And the third one is polishing the writing.
• 22:54
  SPEAKER_01: For example, English is not my first language.
• 22:57
  SPEAKER_01: None of us.
• 22:58
  SPEAKER_01: So English is not my first language.
• 23:01
  And you don't want to submit a paper to a journal with some
• 23:07
  clumsy phrases or sentences.
• 23:10
  SPEAKER_02: Well, you don't want to, but I often do.
• 23:12
  SPEAKER_01: And...
• 23:14
  Because sometimes English will make them doubt your capability
• 23:19
  as a researcher.
• 23:20
  SPEAKER_01: Yes, absolutely.
• 23:22
  SPEAKER_01: And when you have it, it can help you polish your...
• 23:26
  SPEAKER_01: your sentences,
• 23:27
  SPEAKER_01: the phrases and yeah,
• 23:29
  SPEAKER_01: that first it can help you gather the literature,
• 23:34
  SPEAKER_01: the paper that you need and the roadmap to your paper.
• 23:38
  It could also help you organize your literature and also have
• 23:46
  insight into
• 23:47
  different papers even before you delve deeper into each of them
• 23:52
  and later can help
• 23:53
  SPEAKER_01: you polish your writing.
• 23:55
  SPEAKER_02: Yes.
• 23:56
  SPEAKER_02: So thank you for coming.
• 23:59
  SPEAKER_02: We will have you again to talk about how you use AI in teaching.
• 24:04
  SPEAKER_00: I agree.
• 24:04
  SPEAKER_00: Yeah.
• 24:05
  Because we're really interested,
• 24:07
  but it's,
• 24:08
  we've got,
• 24:08
  you know,
• 24:09
  20 minutes or so.
• 24:10
  So we'll do that separately because I think teachers are
• 24:14
  excited to know how other
• 24:15
  SPEAKER_02: teachers are using AI and that'll be fantastic.
• 24:20
  SPEAKER_00: Thank you so much.
• 24:21
  SPEAKER_01: Thank you for inviting me.
• 24:23
  SPEAKER_01: And it was a pleasure to come here and talk about it.
• 24:26
  SPEAKER_02: All right.
• 24:27
  SPEAKER_02: And you'll help us spread the news of the podcast around the world.
• 24:30
  SPEAKER_01: No problem.
• 24:32
  SPEAKER_02: Thank you.
• 24:32
  SPEAKER_02: All right.
• 24:34
  SPEAKER_02: All right.
• 24:34
  SPEAKER_02: Bye.
• 24:35
  SPEAKER_02: And thank you.
• 24:35
  SPEAKER_02: We'll see you next episode.
• 24:36
  SPEAKER_02: Thank you.