Skip to navigation Skip to content

TRANSCRIPT: Tales4Teaching ep. 66 – From information overload to clarity: a student’s perspective on ChatGPT in research

Transcripts are generated using a combination of speech recognition software and human transcribers and may contain errors. Please check the corresponding audio before quoting in print.

Intro: Welcome to Tales4Teaching, a podcast where we explore stories with purpose in higher education. We will share expert insights, engaging interviews, and thought-provoking discussions that will inspire your teaching.

Joan: On behalf of Deakin University, I would like to acknowledge the Traditional Custodians of the unceded land and waterways on which you are located. I acknowledge the Wadawurrung people of the Kulin Nation as the Traditional Owners on which this podcast was recorded, and I pay my respects to elders past, present, and future. My name is Joan Sutherland and this is Tales4Teaching brought to you by Deakin Learning Futures. Hello and welcome to today’s episode where I’m excited to be talking to Supriya, a Deakin University student, to discuss how generative AI is changing the face of her research from a student’s perspective. And she’ll be sharing her experiences on generative AI and how she believes it can support students to transform their experience. Hello and thank you for joining us,

Supriya. It’s great to have you on the podcast today.

Supriya: Thank you so much, Joan. Thank you very much for having me on and it’s great to be having this conversation with yourself as well.

Joan: Can you just tell us a bit about yourself and you’ll interest in AI.

Supriya: Sure. So, I effectively started out with an interest in University. I did my bachelor’s in mechatronics engineering, at the time Robotics was quite a trend and a new and upcoming thing, I decided to take on AI after a few years in a closely-related field to what I had studied which was manufacturing. And so effectively then I realized if there’s any way to work on an actual robot, it would be through AI, I think. Yeah, that is pretty good in itself, considering that it’s literally the processing unit of the robot.

Joan: So you’re currently studying a master’s in AI, is that correct?

Supriya: Yes. I was studying and working up until just recently as well. So, there were quite different in terms of the types of work, I suppose. In terms of studies, it’s been far more sort of cutting edge and it’s been really interesting.

Joan: How long ago did you start your master’s?

Supriya: I started back in 2020, just in time of COVID.

Joan: So, a few years ago now. So how’s your understanding of generative AI evolved since you started your studies?

Supriya: Well, I guess I haven’t worked as much with generative AI. My degree has oriented around machine learning, deep learning, and finally, peeking at reinforcement learning, which is considered to be the most abstract in terms of the kinds of algorithms and applications.

Joan: And I believe you’ve been part of, well you launched the Deakin AI Society. Is that correct?

Supriya: Yeah, absolutely. So, we just launched last year. We realized there wasn’t a platform at the time for students to discuss, and I thought it would be, that would be a great opportunity for everyone to get together, especially the online community, which most students studying the current program in AI are currently comprised of online or correspondence students. So it would’ve been great to have a platform where we could all meet and collaborate. The other, of course, was to actually delve more into the student perspective on AI and all things happening in this field.

Joan: That’s great because that’s what I want, I want to get from you today. So, it’s great that you launched, I know you co-launched it with some other people as well, to bring people together for AI. So, in higher education in particular, there’s been a huge shift in the discussion and the focus and the use of AI tools such as ChatGPT. I’m just wondering how are you leveraging it from a student’s perspective, if at all?

Supriya: Yeah, absolutely. So I think ChatGPT has been tremendously useful ever since it’s launched. A game changer really, has for me, at least, helped me to speed up research, particularly answering some of the more basic and sometimes even more advanced questions. Providing e.g. a Preliminary structure for when I’m writing a passage. Yeah, it’s been really helpful.

Joan: So you’re using a ChatGPT in particular to assist you in some writing tasks rather than doing it for you. So how is that deepening your understanding? How is it helping you to deepen your understanding of key concepts? I suppose.

Supriya: That’s such a great way to look at it. I think because yeah, there’s a couple of fundamental ways in which ChatGPT has supported my learning and research, which is via understanding the concepts faster and also sort of being this mechanism for consolidation, this platform that really helps to pretty much mine every single relevant information there is online about a topic and really condense it, which speeds up research time by a lot while also providing a comprehensive picture. So I’m almost always, it always helps that to know and to have that confidence that there isn’t any information that I should know that has been missed out. Which up until I feel hasn’t really been part of our research was conducted. And I would love to delve more into that in relation to how research is conducted at the moment and the kind of limitations I’m often encountering in this space. And then it would also be great to speak around the kind of benefits and learnings in terms of language, in terms of written communication skills that ChatGPT also provides. I think yeah, those are the two fundamental ways in which it has really supported my learning.

Joan: And there’s a lot of discussion around the validity of it and using critical thinking and evaluative judgment when actually using it. How are you implementing those skills from a student’s perspective, when using tools such as ChatGPT?

Supriya: ChatGPT doesn’t really take away from critical thinking. It augments it. I feel because we need to have a certain amount of information and as you said, depth of knowledge before we can sort of draw patterns or add to it or, or, or think about an issue with understanding of all sides. And I for one, you know, for sure, am the sort of person who does prefer to think through things by themselves first. So that gives me more of a, it’s, it’s, it’s I guess I suppose it’s related to just having that sort of that ability to problem-solve through something and being authentic and original for me, and I think a lot of people, is really important rather than rote learning or copying off someone else’s work, or relying too much on another system and not utilizing one’s own cognitive skills. So, I think, I think researchers, particularly I feel the research community could really benefit with I think addressing some of the ways to not just do research, but how to conduct research.

Joan: Absolutely. Can you talk, can you talk a bit about the models that you talked about earlier? So you talked about deep learning, reinforcement learning, and machine learning. So how does, how does that type of learning inform the AI that you’re using?

Supriya: Yes, So ChatGPT is based on a language model which is based around this very sort of up and coming auto transformers as you know, which has really revolutionized the AI field. 

Joan: How does your knowledge of the underpinning models that inform the likes of ChatGPT, change how you use it or inform how you use it? And where I’m going with that is there’s a lot of discussion around prompting how you actually prompt in and the validity of the results that come back. So is it true, is it not true? How do I evaluate that information? So are you more sceptical of the information that comes back or do you have more of an understanding that hang on. This is really valid information. This is how I’m gonna deal with the information that comes back at me.

Supriya: Yeah, for sure. So I think that can be quite challenging to pick up some times. Sometimes it’s rather obvious. But when it comes to e.g. looking at an example of code implementation, it can be more challenging to pick up. It’s only after, when, once we’ve implemented it or tested it against other similar types of results from GitHub or other code platforms that we can truly discern that. But for now, I think there’s a lot of repetition as well in the answers returned. And GPT4 for is, it’s sort of better at returning certain, it’s a bit more nuanced, but still, yeah.

Joan: It sounds like you don’t need the underpinning knowledge of the language models to look at the output. It’s more about just looking at the output using evaluative judgment and saying, asking yourself, what is this actually coming back with?

Supriya: I think it’s going to be generally helpful across the board. Yeah.

Joan: If it’s gonna be helpful, how do you think students can best leverage AI to support them in their studies? Because there is a real concern around the plagiarism side of things. Yes. That is a tricky topic to navigate around. We probably may need to reframe how we think about plagiarism. In the up-and-coming years, because most students already have access to a vast library on themselves. And I think that in itself could be seen as a form of, I suppose, plagiarism. But it’s really about how these tools help us out in the real-world. If there’s any real downsides to utilizing these tools, such as with neuroplasticity, plasticity, or other sorts of areas. But I don’t think that necessarily utilizing these tools for a test or an assignment is bad. So long as people, as you mentioned, utilize that critical thinking faculties to build on top of these results. And perhaps even somewhat not rely too much on that instant feedback. Maybe, maybe thinking through things and noting down their thoughts before they, they search up anything would probably be the best way to go about it, maybe.

Joan: Well, that’s one thought. And then you mentioned around industry partners and understanding what’s actually in the industry focus from a student perspective, would you like to see within courses to say this is being used in the industry and this is how it’s being used. And I’m thinking more because this is moving so quickly. So, what’s relevant now in four years’ time may not be relevant. And your example of robotics was very, very true, very reflective of this as well in a, in a, in a sense. Do you think embedding the AI throughout the curriculum is the way to go? And how would we do that effectively?

Supriya: Oh, absolutely, that’s such a great way. That’s actually such a great point, I think. Yeah, learning how to do effective research and conduct analysis is in itself a huge skill to have and it’s critical to have that. And I feel, I honestly feel like with IT and engineering should be augmented, not just with, I guess AI, but also with other fields that help them think, such as humanities and all the other sorts of fields which really question the sort of deeper understanding of things from other perspectives. So I think that it’s really important to have that holistic education of not just doing, but also thinking about how to do something.

Joan: I love that perspective about embedding within the humanities as well, the ability to think, because our ability to research and analyse you’ve hit the nail on the head there. It is such a critical skill to have and one that we don’t, we can tap into, but you can choose not to tap into it as well. It is a skill in itself. So the way to think and how do you think and how do you question. And there’s a lot of discussion around creative, creative thinking and different ways of thinking. And they’re going to be the next skills that people are going to need. Because we’ve got this, these different tools in industry already and some of these tools have been around for years. How do you actually create industry ready graduates with the appropriate skill sets? And one of those things is those thinking skills which are deeply rooted in the humanities, but there’s no reason why they shouldn’t be embedded within all the other industries as well. So great point there. Thank you. 

Joan: So what are the ethical considerations do you think are important in higher education and how can we ensure these technologies are used responsibly for a student’s perspective?

Supriya: Sure, thanks Joan. That’s a great question. One that comes across very often in discussions around AI. So, I think one of the more general concerns at the moment is around bias and discrimination removing those aspects from AI models there is a strong push towards debiasing AI systems towards different demographics however I think it’s important to keep in mind how that actually plays out in real life. In fact some of the examples that have come out recently from ChatGPT do not apply the debiasing approach, the same approach to every group which you know which is not true truly ethical system I would think perhaps and sometimes also the models are neutralised to such an extent that there is a blanket approach to even the existing realities so that perhaps could be a discussion around that but in terms of academia I think noted earlier in our conversation around plagiarism and how GPT is employed by students definitely I think there can be strategies to help individualise what students are working on to reflect more of their contributions in a more creative and critical way rather than removing AI models and any digital tools as a whole.

Joan: And so to clarify, and it’s an important point you make in relation to de-biasing. So when you’re talking about de-biasing AI, making sure that the AI isn’t unfairly treating some people differently to others. So it’s really important, especially in relation to that minority groups that you’ve talked about. And there’s a lot of great work to understand the data that is AI is trained on. And it’s important to understand that this can be biased or can have prejudice. And therefore, it can inherently trait groups unfairly because of these so to de-bias it, we need to identify the unfair parts essentially and fix them. However, in the likes of ChatGPT it isn’t so easy, which is where critical thinking skills, and evaluative judgment on the individual level needs to come in to be used in conjunction with AI. AI doesn’t currently have the human values like whether or not that will come in. It doesn’t have the ability to judge a situation that doesn’t have the ability to take into account the individuals as you were saying, the bias or the cultural considerations, anything like that, where it is unique to the human versus the algorithm itself. It’s great that you’re thinking about these things, especially from a student perspective and it’s really insightful. So I really want to thank you for your time today and I look forward to talking to you more about this in the future.

Supriya: Yeah, absolutely, Joan, thank you so much. And yeah, it’s been a really great discussion.

26 April 2023

Last modified: 29 May 2023 at 12:53 pm

back to top