TRANSCRIPT: Tales4Teaching ep. 67 – AI: revolutionising industries and transforming education

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.

So today I’m joined by Jesse McMeikan, who is the manager of the SIT Innovation Group here at Deakin University. Hi, Jesse! Welcome!

Jesse: G’day, Joan. Thanks for having me. Yeah, looking forward to having a chat.

Joan: Well, today we’re going to be talking about the relevance of AI in industry, and how it’s actually transforming education. So, let’s get started. Can you just tell us a bit about yourself and your role at Deakin University?

Jesse: Yeah, sure. So, I’m the manager of Industry Projects within the SIT innovation group. So, we’re a professional team within the School of IT, and we predominantly manage the final year sort of common call industry capstone program. So, 10 different disciplines all come together in the final year, and they basically spend two trimesters working on building products, games, apps, data science experiments – you name it. And so, they do that within six school companies that focus on our areas in the school.

And in addition to that, our group also manages the innovation agenda for students within the School of IT. So, things like hackathons, boot camps, competitions that we often go in, and things like rapid prototyping and startup ideation, that sort of stuff as well.

Joan: Okay, so technology is right at the forefront of what you do and making sure it’s relevant to students.

Jesse: Yeah, we’re really lucky in the sense that we get the students coming in in the final year. They’ve gone through the sort of process of learning and the different discipline areas. And so, in our program, they get to put it into practice, and it’s very much sort of a black box. There’s a lot of autonomy and freedom that we give the students. And so, a lot of the time we see students experimenting with new technologies and things that are just emerging. And so, it’s fun for us. It’s also fun for the industry, people that come in and collaborate with our students.

So yeah, things like what we’re talking about today. AI and other emerging areas are starting to really sort of show up in in what we’re doing.

Joan: So, ChatGPT obviously is the biggest one that’s taking the world by storm. And now we’re talking a lot about it in higher education. Can you give some examples of industries that are currently using AI, and what benefits they’re experiencing from the implementation.

Jesse: Yeah. So, we’ve had some relationships over the last couple of years with different sort of startups or other organisations that have been interested in sort of collaborating with this, and we’ve started to see what they were doing. There’s obviously, as you said, been an explosion with ChatGPT that’s emerged at the moment. But you know, before that we were sort of seeing things around like computer vision. So, some startups that we are working with looking at interesting ways of using computer vision to identify, you know, H-VAC. So, you know, like heating ventilation, air conditioning type stuff.

So, they would go in with the 360 camera, and they would use that to record all the different elements within a room, and they were working with our students a couple of years ago on that, using algorithms to sort of detect it. And you know, from an efficiency point of view, that was instead of someone with a clipboard having to identify and write everything down, they were literally just running through, scanning it and then processing that vision later.

And so, you can imagine the efficiency that they were getting from that, and our students were working with that company. I think they were called Older One. We’ve also seen, which is quite popular with our students, sort of the emergence of chat bots. And this is pre-GPT. But you know, chat bots were emerging in terms of like Help Desk type stuff, quite, quite basic. But that functionality has been there for a while, and so that was a slowly emerging we were seeing, you know, and it’s obviously focused on like efficiency, right and boosting productivity.

But you know now, with the emergence of ChatGPT, and basically the GPT stack that is starting to emerge, API keys which give you the ability for you to program and build your own applications that then can interface with it directly, there’s just an explosion of development, particularly in the open-source community that’s emerging. And so, you know I’m expecting to see, well I’m already seeing it, if you keep your ear close to the Twitterverse, you know call centres, for example, are going to be one of the first that’s probably going to be massively impacted by this. You could probably say things like, what was interesting to me was game development like 3D artists, game developers, they’re using things like this Mid Journey Version 5, which is just come out, and there’s also Stable Diffusion. So these are sort of like generative AI’s that can actually create images. So, what we were saying was like, you know, or in 3D art design 3D animators, the work of 10 animators in a startup, you know, is now down to like two or three, and the ones that are keeping their jobs are the ones that are able to effectively use these new tools.

Yeah, it’s interesting, but it’s yeah, it says things like the timelines are shrinking in terms of what they can create. I saw something in the other day that was fascinating. It was a clip on Twitter, it was a 20 year old game Vampire: The Masquerade which came out like 20 years ago, with very basic graphics. Someone had run it through Stable Diffusion and now it’s photo realistic. You see these games that are sort of emerging, I think it’s going to be a whole like redux cottage industry that emerges in film and TV, 3D, all that kind of stuff. Fascinating.

Joan: Well, you mentioned about the animators, for instance, that have reduced down to two or three within a startup. Now, is it about working alongside the AI, I suppose and taking that on board? How do people, there’s a lot of threats out there in the sense of threats to your job and you’ve mentioned around creating efficiencies. And that’s something that AI does. How do you recommend that people in the Education Community encourage people to work with versus work against different AI?

Jesse: Yeah. Yeah, it’s a really good question. So where when I saw this example come through about this 3D sort of game team. It was an Indie developer, and the person was quite unhappy, right? They were complaining about how like, you know, this is not what I signed up for, and you know the boss is keeping the person that’s using the tools, and I’m possibly gonna lose my job. But I think you need to look at it differently, and it’s a lot of what we try to like, talk to our students in the capstone program about as well, and the school, is that you need to be able to think innovatively, right?

The people that are gonna survive in this space, people that are just gonna expect to have a cushy 9–5 job, or work for someone, you’re gonna have to actually think, ‘How can I use these tools?’

So, you would expect to see more games get developed, right? Because there’s more ideas, and there’s more accessibility. There’s more ability for people to actually do this stuff now, in a lower budget. You know it’s interesting, that point around the budgets like VC runways a venture capital runway, that’s a big conversation that’s happening at the moment, right? Because when a venture capitalist comes in and give seed funding and angel investing for a startup idea, they would factor in the cost that it would take for, you know, one year, two years of runway. And now that thinking has actually come right down. So you don’t need a 10 million dollars investment or a 1 million dollar investment you might only need a hundred thousand investment.

Joan: which is much more achievable.

Joan: So does this excite your students, or does it terrify them?

Jesse: Yeah, I look. It’s funny, right like I think some of them are really keyed into this stuff and dialled in. I think many of them aren’t, which is surprising. This is going to be huge in the IT industry and our industry, and it’s gonna affect them particularly if the young people haven’t started work yet, and they’re entering the workforce, this is gonna be part of their life in the same way that we don’t do sort of. You know, we scientific calculators and things now to do, you know, a lot of complex maths.

This is gonna be part of their life. They’re gonna have AI present with them, so I don’t think so. I think we need to talk about it more, but we also, it’s just moving so fast like, Where do you take it out? Like, what do you tell them?

Joan: Yeah, 100% It’s going at the speed of lighting. It’s just absolutely crazy as things come out, and as you mentioned before, on Twitter there’s always something new, someone’s got new ideas, new perspective, new resources coming out, everything, and it’s developing daily.

Jesse: It is, it is, and you know the thing is to like, if we’re talking about ChatGPT in May 2023, we’re way behind the curve, right? It’s not about ChatGPT anymore. You know people are now talking about AutoGPT, I don’t know if you’ve heard of this AutoGPT and BabyAGI. All these sort of open source projects are emerging on Github, getting massive numbers of stars and these are kind of like self-directed, self-prompting AI bots, basically. So, rather than having a single ChatGPT conversation or AI liaison, you’ve got a 100 working consecutively towards the goal. That’s pretty fascinating to watch.

Joan: It is fascinating and terrifying. I’ll use that word again. It’s funny that you mentioned that around the auto agents. So that’s around the ChatGPT or the large language models behind that that requires a human input to respond to it, whereas the auto agents that you’re referring to, that’s generating its own response. Is that correct?

Jesse: That’s right? So this is a program that you can, just Google it, you can Google AutoGPT. But yeah, you can basically give the AI agent. A high-level goal, it may be, you know, create an effective business that you know makes money year on year, or whatever it will identify like five goals that it needs to do that, and it will just start running. It will just start like researching, it can browse the Internet, it can write code, it can output things like files, create files, do all sorts of stuff. It’s pretty wild.

Joan: It’s very wild. So how did you say you touched on it a bit that your students need to be adopting this, and really being partnering with AI. ChatGPT, although a lot of the buzz is around that, AI is a massive industry in itself. So how do you see the role of AI evolving in the future? And what industries do you think will benefit most from its continued development?

Jesse: Look, I think that’s like the 1-million-dollar question, right? I think almost every industry is going to be impacted, in fact, by it, and I think almost any industry is going to be able to benefit from it. I think I think new industries may emerge. I talked about like the cottage industry around like reduxing things, you know film industry, right? You got to be able to go back and add, remove characters, and, you know, change scenes and things like that. I guess the question is, if you’re running a business now, if you’re any like a small to medium enterprise business that may be information related or services related, how’s your business going to function with an employee that has the knowledge depth of 10 million PhDs, right? And can synthesize and reason and solve complex problems for you, and then can actually create things like artifacts, can actually like publish text or update your website or answer queries.

You know we haven’t really encountered this sort of thing before, like this is, this is all really new, and I think kind of any industry, maybe that has, like large-scale complex coordination and maybe has clear data imports is going to be an industry that’s going to benefit from this, because just the ability for, like an AI agent to be able to compute and calculate things at such a broad scale and taking so many different information inputs, is gonna it’s gonna be like a really huge productivity dividend for those sort of businesses. So maybe a transport logistics is probably one. We’re not quite there yet, but like, maybe like large scale software development? You know, I’ve been playing around with how can you get AutoGPT to play the role of like a 10 person scrum team. So this is this, I’ll be like trying to try to basically get it to play the role of 10 different people in a 10 person scrum team with a goal, right and see like, can it actually discuss things in issues you know, using scrum methodology, because there’s a lot of content out there that it can learn about this. And so it’s been some interesting results.

Joan: How did it go?

Jesse: So what I found with these auto agents is, it’s about how you prompt them and ask them to do different things. So, you can sometimes get stuck in a loop where you ask it to research and understand something.

So if part of what you’re asking to do is like research, understand scrum, and then, like, apply it to this, it will learn about scrum, and it will try and do something, and then it will come back to learning, and then it will do something that will come back to learning, and it can sometimes just get stuck in this like endless like cycle of trying to learn, learn, learn, learn. It’s better, I found to be able to like directly ask it to do something  and then it goes and like figures out what it needs to know. That’s the bit that we’re going to have to figure out ourselves, I think, is like, how do we actually use this effectively?

Joan: That’s probably why the industry of prompt engineering has popped up as well like, actually create to get something to do what you want it to do.

Jesse: That’s right. This is just learning. And, as you said, we’re going to t it. That’s right. What’s interesting? You say that because one of the more more interesting things I’ve come across is this idea of a super prompt right? So I’ve seen some people that have started sharing their super prompts, and the idea is, the AI knows how to prompt itself better than you will ever know how to prompt it.

So, create a question, create a prompt that basically gets it to go through and come up with the best prompts to achieve the goal that it needs. So it know more about itself than you ever will, so we kind of almost have to remove our puny human ego from this and basically bow down to the AI and basically ask it to figure out things for itself.

Joan: Well, one of the big questions I know we’ve been grappling with is around, there’s been a lot of discussion around assessment and changing assessment so that AI actually can’t do it. I suppose that we’re assessing, learning. But the other question or the other thing that’s really come to the forefront is around that need for critical evaluation or evaluative judgment and critical thinking skills as to what you’re actually getting back because I suppose you can put these prompts in and super prompt, and they’re going to know so much more than you’re ever going to know about. How do you know what you’re actually getting back is relevant for you? That’s authentic, you know, and something that you can apply to your own context, I suppose. What are your thoughts around that?

Jesse: So, you talking about the into the integration of AI into assessment? I think, Look, I think the thing about like how we think about this a in terms of how we how we use this with students like, I think we we’re going to have to really redefine how we assess? Because is it that we’re assessing students on their knowledge content, right? And what they know, when you got this tool that knows everything right that you can just pull things out of? Or are we assessing them on like the outcomes that they can achieve right? Because like that’s sort of the thing like, I see in some of the things that we’ve got coming through in our program like miraculously this year overnight. Seems that everybody is really good at writing summaries and documents and documentation, which honestly was a huge headache in the past, because I didn’t do that very well. And now they do it really well. And so now everybody understands the documentation. They can read it.Am I happy with the outcome that’s been achieved. Yes. Am I necessarily happy that they know everything that they’re doing? I don’t know. I’m not sure. But then the question is like. If these things are going to get better and better, better, do they really need to know all that stuff now? Yeah. But then, like, what are we doing then?

So we have to think about like, what is our role in this? And you nailed it when you’re talking about like critical thinking and problem solving, because these things still make mistakes, right? You make mistakes, and you can see, I can see some mistakes that it’d be made in this some of copy paste Haven’t really thought through. So, they’re going to get pinged on that. It’s a big conversation.

We could, clearly we could clearly apply these tools to what we do as well, right? Academic workload, for example, is a massive, massive conversation in our school, as it is in other schools. We predominantly now, you know, even before COVID, so much more now, everything is online. So everything is information based, right? We all use these different learning management systems which are basically popping out information, we use contract a lot, but feasibly. You could see that, like we could discuss the idea of using ChatGPT to assess and mark assessment, write feedback because you can clearly like, tell what you want it to be marking. It will do that now. I guess the question that is like. Would we accept this? And would students accept this?

I don’t think they would, or at least not right now. I don’t think they’re prepared to accept that right now that they’re being marked by AI, even if that AI is going to give them better feedback, more attention, be more engaged. Maybe.

Joan: It’s a value-add of university as well though why you actually go to higher education institutions, and who you’re actually coming across, industry leaders, people that are actually teaching you, and you’re learning with as well. So that’s the other value add of higher education versus if it was just knowledge, you could get a lot of Youtube, for instance. Right? It’s yeah, it’s more around the experience. And there’s a lot of different components that add to that experience.

Jesse: Yeah, you you’ve touched on like a really interesting point, right? Because now it comes down to well, what is the value proposition of a university? And we are known in Australia as being quite a digitally savvy university, premier digital experience and all this kind of stuff, and actually huge benefit for us when the COVID lockdowns happened, because we just transition to everyone to our Cloud Campus, and we were off running.

The question, though, is if we are going to embrace this technology and embed it into a lot of our teaching, we actually are possibly going to diminish our value proposition to students, right? Because if it’s so easy to do it, everyone’s going to do it.

So, what exactly is your call out to students to be enrolled at university, if you can ask AI, which you can do, you ask ChatGPT to develop a 12 week curriculum on software engineering, it will do it for you, and it will draw it from 10 million different (inaudible), and it’s probably going to be very good. Now, maybe that’s fine if you’re busy and you can’t do university, but like, maybe you want to come on campus and have that experience. Maybe you want to have face to face engagement with people.

So it’s kind of a bit of like do we actually go back to analog? Because that’s the thing that we can offer that they I can’t. I don’t know – big question.

Joan: It’s a big question, and one that I think every institution grappling with, not just higher ed, but, you know, like where it actually fits, because there’s so many challenges with integration of AI from a value proposition, but so many other things that pop up, especially because it is part of our lives now, AI. And then five years in the future every industry is going to have it a lot more prominent, I suppose I would imagine, so how we actually ensuring our students are graduate ready for different industries, which is what you do, I suppose, in your role? Right?

Jesse: Here’ s a thought experiment. What if you took every single piece of assessment, every single piece of learning material, every single thing we’ve ever done at Deakin, then can put on a database somewhere, and you train a large language bot on all of it, and gave it self-directed goals to become a University lecturer with that one AI agent, be able to deliver everything that we offer, and maybe not everything, but like everything that is possibly like information based, it could possibly do it in the future. I don’t know.

Joan: Maybe if it’s information based. But again, it goes back to experience, and my background is in teaching and learning design. I always think about like, I’ve tried the, you know map out the twelve-week course and done a few things around that, and it does. You know if you were starting from scratch and you were thinking, oh, how does this work chronologically, and what works, and what doesn’t, but the power of like an educator being that you know you change things around. You make it more engaging. You can do different things. Use implement different strategies. But, as you said, I don’t know where it’s going, but it’s an interesting thought. Yeah. It’s actually scaring me now. But that’s alright.

Jesse: Well, you know it’s interesting. And it is honestly like, you know the ethical dilemma is there? It’s danger to this. I’ve been talking to people about, you know, certain things that you know, this is out there in the in the public now. So, you see a lot of stuff emerging on Twitter right? What really cool stuff. The thing I’m thinking about is, what is the stuff that’s not being posted on Twitter? People are posting all these really cool, awesome things they’re doing with it, not posting the maybe not so good things that they do that right like you can imagine. The genie is out of the bottle now.

Joan: Buckle up. I want to thank you for your time. But before we go, what advice would you give other institutions around AI to ensure graduates are industry ready?

Jesse: The only thing I think you can do is expose them to this, right? Like you’re actually failing at your job if you’re not exposing students to this sort of stuff. If they’re emerging out of the pipeline, you know they they’re joining a business and they’ve got no idea what’s going on here. And we’re not doing our job.

You know, I find in Australia graduates that are imagining, you know, they look at our students like the Google wizz-kids, you know, like our students kind of want to join companies thinking they’re going to learn all this new stuff. But a lot of the time actually like the newest stuff is coming out of like the universities, right? So they actually have to realize that they can be quite influential when they join, because they’ve got the most hopefully the most current information.

But yeah, I think I think we have to be smart with how we create our units. We have to be flexible with our curriculum. We have to somehow figure out, particularly in IT, at least how we can create units that have the ability to just be refreshed and updated as fast as we can.

Joan: Yeah, it’s constantly changing.

Jesse: And if you’re not in IT, this still gonna affect you, because really, when you think about it, it is the glue that holds a lot of things together. It’s everywhere.

Jesse: And so you know, health for example, this is going to have applications around like fast diagnosis. We’re already seeing that, like X-rays and ultrasound are being analysed by AI agents that are actually doing some pretty amazing stuff and detecting tumors and all sorts of stuff. It’s going to be everywhere. So, you really just have to be open to the possibility that this is going to affect your industry and your students.

I think we’re probably gonna have a better idea in about 10 years where this is going, but at the moment we’re right at the beginning. So it’s gonna happen whether we like it or not.

Joan: and it is happening, and that’s what’s exciting. And I’ll go back to terrifying as well. But it is an exciting time to be part of higher education in particular, and see how it’s transforming people’s just thought process, and what we’re doing, and how we’re doing things, and how we can engage students and make them more industry ready when they graduate.

Jesse: AI ready.

Joan: It’s all about AI. Before, what was it? 2020 was all COVID. And then this year 2023 it was ChatGPT, and now it’s gone to AI.

Jesse: Twenty-first century graduates. I think we’re in the AI age now.

Joan: I love it! I’d like to thank you for your time today, Jessie, it was really enlightening, speaking to you about what you’re encountering in your position, and how students are dealing with AI as Well.

Jesse: no worries. Thanks, John. Thanks for having me. And yeah, really, really interesting conversation that we’ve had. So, you know, I look forward to where this is going to go and what we’re going to do with Deakin.

Joan: Definitely. Thank you.

Jesse: Thanks.