Generative AI P2: Collaborative Inquiry Into AI and Higher Education - First Encounters of an AI Kind

This study was shared on May 4 at Middlesex University Dubai at the 2nd International Conference on Technology, Innovation and Sustainability in Business Management (ICTIS 2023). 

Find more information: https://www.mdx.ac.ae/ictis2023.

AIMS OF STUDY
The aim of this auto-ethnographic study is to provide insight into the lived experiences of higher education faculty in their attempt to design a new educational module with an advertising/marketing specialism making the best use of  Artificial Intelligence (A.I.). This is in light of increasingly prevalent social robots, such as ChatGPT4, Dall-E, Designer etc., which are particularly disruptive within the field of advertising. The acceleration of A.I. applications across various fields of content creation has elevated the importance of a graduate’s self-efficacy in working with A.I. in terms of employability. 

The auto-ethnography will be presented in the form of an evocative narrative and will share feelings, hopes, aesthetic reactions, and moral dispositions.  

As befitting a rigorous institutional auto-ethnography, the study will aim to explore the following research questions: 

• Can AI be used to help develop teaching and learning materials?  
• How should we teach AI to advertising students?  
• What are the most effective and useful programs that infuse AI concepts within an advertising curriculum?  

The authors are both affiliated with Middlesex University Dubai:

• Stephen King, Senior Fellow HEA, Senior Lecturer in Media
• Judhi Prasetyo, PhD Candidate, Senior Lecturer in Computing, Informatics and Engineering

Correspondence via s.king@mdx.ac.ae.

Chapters

• 1:40 Start of Podcast

Transcript

Hi, everyone. Welcome back to our second podcast about AI in higher education. I'm Judhi. 

Steve.

Yes. What's up, Steve. 

Yeah, everything is going great. We are progressing towards our presentation on, may the fourth. And we are ready to share some of the insights, we are only going to have a short time to discuss on Thursday. So some of this content that we will talk about today will fill in those gaps. So we're going to be talking today about the first week of our study. Yeah, it was a multi-week study. And I produced a journal using Microsoft OneNote. And I've captured images. I've t ranscriptions of my voice and I've been recording my feelings and my experiments over the course of that period. And that's what I've shared with you and I'm v ery interested to find out which of those experiences that you felt were most interesting and worth exploring a bit further. 

So last week we have discussed the aim and also like the ethical the aspect of it. And what's the progress so far?. 

We've gone through the whole study. It's complete now. Which is fantastic. And we have managed or, I have managed to produce a full module narrative with aims learning outcomes, with assessments, with the syllabus. That's complete to a very, I think it's a very good standard. I have also explored the production of content for lectures and have managed to get 8 weeks of content produced. And I have five weeks of workshop materials, which are I think are very helpful. So I've got all of that ready. We have a three week diary t o explore and dive into. A nd I'm looking forward to presenting our paper at the conference on thursday, may the fourth as well. 

That's pretty impressive. So I guess you collected a lot of data sources. Yeah. 

So the first step. In the first week was all about preparation. The first week was about reading through the journal articles. Was deciphering what had already been discovered. Unfortunately, a lot of this material has only started to be published in the last year or so. 

And that means that the content that we are r elying upon from the traditional academic sources is probably two to three years older than that. Given that Chat GPT just came about in November, December of last year and it has only begun to be widely known at the beginning of this year. That means, all of that content is Of less value. It's not no value. It's just of a lesser value. And so the panic that I had at the beginning was how do I supplement that with something which has been more practical? And so I would be using podcasts. I found a couple of very strong podcasts. I think they are very strong. Obviously it doesn't go through the same rigor as the academic publishing, but the podcasts have got some value. And I've also unfortunately fell into the TikTok environment. And I've been relying a lot on those types of. Yeah. It's a as a tool. Yeah, that's what I've been exploring initially in terms of the literature preparation. 

Okay. TikTok is not very academic per se. So how do you filter that? Or. 

I love the word filter, because this is all an it's an AI-empowered tool. So as soon as you start browsing TikTok and you start finding one or two stories and start liking it or sharing it or saving it. Then the algorithm starts to recognize that this is the kind of content you want. And through its hashtags. And what have you. It will start sending you more and more of the same content. So within an hour or so, you will end up with 200, 300 videos of pseudo experts who are talking about the subject. And subliminally, you will start to see the same kinds of software that are being recommended, or the different tools and it's very useful for a quick environment scan. Even if it's only very skin deep. It's very useful to giving a direction for deeper research. 

Yeah. Obviously we are not the first one who thought of using, AI to develop a course. But on the other hand, these things are pretty new, a nd the development in the AI is pretty rapid in the past few months. 

Yeah, it's very exciting. It is a furore. It is a frenzy online of pseudo experts who are saying that you can use v arious apps, which we've used in our study and tested to create courses. And that's one of the motivations for this study: "I s this actually true?" 

What can you realistically deliver just from a few prompts into a machine? Is this going to revolutionize education or not? And how is it going to transform it? And that's one of the motivations for this particular study. If we pursue this in a methodological fashion, what comes out of it? Versus just relying on what your favorite Tik TOK influencers telling you. 

Yeah. So after you collect all of this data from these sources, what is your next move to compile this to, to help you develop? 

So the next step is to actually do some familiarization experiments. It's not to go headlong into, let's try to create a curriculum. No, the first step is to play with it. Familiarize, become more comfortable with it. 

If I was to attempt to pursue the task straight away with no familiarity, no ability, no information literacy on these technologies, it would fail. And that would not give us a true result. So I've spent some time doing experiments. I've used Chat GPT in different ways. I've used it to rewrite a lot of my own writing. 

Okay. I've done some experiments where I have played with the text to image functionality. I have created comics. I have created websites. I've curated fantasy worlds. I've used it in play function. I've even tried to use chat GPT to pick football results. 

Okay. So it's not directly you use it immediately. To develop your course, but it's for everything, every aspect of your life?

I'm just trying to familiarize and socialize it. So that I don't go headlong in make a mistake, get frustrated and say it doesn't work. 

Yeah. 

I wanted to see, okay, this is what it does. This is what it could do. This is i n a reasonable way without using textbooks or without this is how anyone would this is how anyone I expect would grow. Start tackling some of the more dull parts of my life. I was trying to see whether it enhances the passions in my life. And see where that goes. And that was a part of the first step. I also used the Descript to create a voice clone,which is perhaps the most sophisticated thing I've done. And I made some purchases. 

Okay. 

I made three purchases. 

Oh, wow. So you actually spending money. 

I have spent money ...

So not just using The free tools, but...

So the first thing that I spent on was Descript. I got the premium version of Descript and that allowed me to clone my voice. Which is very interesting because I wanted to be able to do that. I thought that was quite cool. And I could see lots of different uses on that. Podcasting is a bit of a passion of mine. And this, I felt that this was something I would want to explore. The second thing I paid for was a SlidesAI.io. IO and this is something which is about content creation. It allows me to take text and turn it into slides. Now creating slides is one of the most burdensome tasks we have as an educator. 

And I was trying to work out if this could create slides for us, then that is a valuable tool. So I decided to invest in that because that would be, I think it would be a game changer if we can get slides created. And then the third tool was a website hosting service. 

Okay. 

Not directly related to AI, but the content I was creating was so beautiful. And I wanted to host it somewhere. And therefore the correct site for hosting required a certain amount of subscription. So the AI content needed to be seen. And so I paid for somewhere for the AI content to be seen. 

It's like Instagram paid followers? 

No, this is a proper website. It's a proper website. It's called obsidian portal. So it's a desktop role-playing game portal. It's where you create fantasy worlds. I can show you. You definitely have a look at how much effort I've put in and how much content I was able to create in a matter of two days. Yeah, I think it was phenomenal. 

All right. Okay. So you played with all those tools, obviously you're impressed and you went all in with this. But, overall how satisfied are you with them? 

At this point. I was very excited. 

Okay. 

I was very excited. 

Yeah. 

And I was becoming obsessed. 

Okay. 

It was really I think it's important that we talk about now about how my mental state was at this point. I was increasingly showing obsessive behavior. I was addicted to it. And I could see a I everywhere. I was it's part of there's this, the study methodology is the fact that you are journaling and you're going to be put yourself out there more than a normal person would do. So this is part of the process that we've already explored and prepared for. 

But I was not prepared for just how much anxiety that I would start to feel as a result of being continuously involved and observing how AI impacted my life. I couldn't switch off. It was everywhere. 

Okay. 

The AI started to appear in all my social media feeds. It started not just on TikTok, but on YouTube and on Instagram, I started getting new invitations to events and I started going to these events. I thought I had to go. And so I'd go to an event with AI. I'd go to an event which was not AI related. And someone would make an off the cuff remark about AI. 

Okay, so just the catch word?

The people would say, ah, my speech is not written by GPT. It's natural speech. You know? 

They would make a joke and people. And I would hear people make little chuckles and it would just "AI is here. It's following me". 

Okay. 

I would start to see in cafes and restaurants and I would notice changes in automation. "The AI is coming to get us!" My LinkedIn feed I'd see videos of how it was is coming. 

How robots were being developed in Japan and places like this, I would be like, that's not good. The news in this period was enormously negative towards AI. Elon Musk, Italy. There was on Quora. I have a Quora feed was talking about how AI is being used in military and how it's being used in wars. 

It was a really immersive, negative, toxic environment where I couldn't get out off. And it really increased my anxiety to very physical symptoms where I couldn't sleep. Yeah. My temper was shortening. I was getting into more arguments. I was feeling. To the point where I started posting on LinkedIn - about AI - and about the dangers of AI and trying to raise a debate. I was trying to raise awareness and make people you know, share my fear. 

Preach about? 

Preach, preach about it. And also say, Hey warn people I was really scared. And I had, fortunately we have it in the protocol and we followed the protocol, which says like, when you reach this stage, take a day off. 

And I gave a day off. I don't think if we hadn't done that process, I would not be aware of the need to have a day off. And because there was. I would might've felt oh, I was cheating when you, as the co-researcher. I'm not journaling today, but the fact is I needed a day off and I was able to take a day off. But even in that day off... 

…it was the first time, the first day off was I still was seeing AI everywhere. It was still thinking there. I think it was only when we had our first sit-down, or one of these discussions. 

Yeah. 

Which I really valued was the fact that you were able to talk me down. Otherwise, I was at the point where I was thinking we are in a computer. I, there was a point where I was thinking we could be in a computer simulation. 

It's it's a normal behavior that, for example, if you drive a red car, We would see suddenly, oh, there's so many red cars on the road. The next time you drive a yellow car, you will see so many yellow cars on the road. So because now you're studying about AI, then you start seeing more AI around you. Yeah. How does that work?. How does this help you in your goal in making this course? 

So I, at this point now I'm confident. Okay. My self efficacy is up through the roof. When we first started this, we made mistakes on Descript. I couldn't even work the microphone or it was confused. There's a lot of the, if you don't know how to, if you haven't got the confidence to you use the technology, then you're not going to do it very well. So I've worked out prompts that was important. I'd worked out which tools to use. All right. So that was very efficient. So the first thing I asked GPT, what are the steps I need to do? Okay. To do this project. And it came up with some steps and then I found a. There's an AI project management tool called Asana. That's what we use in the industry . 

Yeah. And so I took a free subscription on this. And so I use the instructions, from Chat GPT, plug it into Asana to set myself on with a plan moving forward. Okay. 

Yep. So how many weeks? Since the first time?

Yeah, I didn't look back at Asana again. I set it up the first time...

Not just the Asana but the overall? 

The whole project has taken three weeks, three weeks. 

So one week of prep. Okay. One week of let's say…

Exploration? 

A little bit yeah, continue experiments. And then towards the end of the second week was when I went Yella! Let's go! Let's start! I was confident enough. So if you was so the whole module narrative, which is the first thing I created and the Asana was done in a full day, but it didn't take me a full day. It was only a couple of hours of that day. Okay. Our day's work is in this project is four hours. Cause we obviously have other things going on. Yeah. So in a four hours period, I created not one. But multiple module narratives. 

Usually you haven't even finished one.

Usually it would take you a week for one person to work on it, and then it would have discussions from other people. So I was able to produce three of what I consider it to be. I usually receive these, I don't generally produce it, but from what I have received: these are very good module narratives. I'm very happy with it. With the kind of content that's in there. And we'll be able to share some of those. I'm happy to share the text because it's purely from AI. It's nothing, which is not my real IP. It's come from the software. We can show that. So from a top level of what should be in the course. Yeah. 

It was very good. Yeah. And that didn't take much time at all. The next step was, I was then enthusiastic, but I took a break at this point. And then I went into the next week. This was when I started really with a fresh eyes, to start to produce the content. 

Okay. So I understood that you are very satisfied with the result. In using the AI to design courses. At this point, right? But it seems that you have some reservation. At which point that you see that, okay, this might not be a hundred percent? 

There are two things that happened. And this is in the experimental phase. It was when I tried to use it to predict the football. Now, this might sound silly. But I wanted it to tell me based on trends. What would be the football results this weekend? And it wasn't the fact that it gave me any wrong answers. It just refused to give me any answers. Okay. It refused to provide me a response. Okay. 

And this was the first time that the AI h ad denied me. Okay. If you get me. Before that it was doing everything I wanted. Everything was given. Okay. Wait, it might not been very good. Yeah, but it was doing something this was saying, no, I am an AI. I do not understand football. And I was like what use are you? 

It's the same thing. If we ask AI to predict the election result for example, it will give just the same thing. I think I don't know whether you feel the same. Yeah. AI is good to produce something if we expect we know what to expect. Like we know roughly, this will be the result. It's just that we are too lazy to go through that process. And we get something to do it for us at the end, we will read it. And yes, this is what I want, but prediction is different things. So in your case, I think this worked very well because the domain very well. The education. The curriculum. What does he take? All these modules, all the subjects. But we just, okay. Let me rephrase, not too lazy, but, it will take a lot of time. 

I like, what you're saying is that there's a couple of elements in here. If I went back to that one is expectations and then there's another element of cognitive bias where I don't see. The amount of effort that I'm actually putting into the machine. And so my initial results were because what I was putting in was quite significant. Yeah. And so I was getting good results back. But I was perhaps imposter syndrome is coming into this in a way I wasn't realizing how good my content was when I put it into the machine. Yeah. And when it comes out it has magnified, that's good. 

But when I'm asking it to produce based on no input, it's it's I'm not seeing the fact that it is my lack of input that is contributing to the failure of the machine to give me a good output. And then my expectations because of the early successes have been so amazing. If you put good work in it, magnifies it to a certain degree. It's not just two times, three , four or five times above your expectation. Yeah. Maybe? T hen when you, when it doesn't achieve four or five times your expectations, your disappointment is four or five times magnified. I think that. That was my first indication that it's not going to be great. There is a second mistake that it made and which was, and continues to disappoint me. And that is in text to image. 

Okay. 

So I was creating social media posts. 

Yeah. 

And it was for Ramadan greeting. 

Yeah. 

I was, I came up with a, quite an interesting image. Yeah. I thought, oh, this looks nice until I zoomed in. Yeah. I, when I zoomed in, they were, I didn't understand what was, it was demonstrate what it was showing, but it is possible that someone might have taken a - a nother meaning. From what the image had shown. 

Okay. 

It had a gentleman and a lady. And if you closed in closely, the faces were not smiling or it did not look like they were smiling. It looked like there was something not very happy going on. And I was like, ah, that would have been a really 

Artwork or visualisation? 

It was a graphic animation. Pure graphic design, no photography. . And I was like, if I had used that without thinking. Yeah, that would have been potentially a big problem. Yes. And as we go, as I went through some of the tests and some of the play testing, I noticed that the text to image does come up with inappropriate. And I'll say that in the broadest sense. Imagery. In fact, the Canva text-to-image has a caveat on it where it refuses to give you images if it thinks it could come up with something, which is, you're not expecting. Okay. So the text to image thing. I was, that was my second big worry. And I continue to think that's a problem. 

So I think expectation plays a huge role here. Because, yes, we would get disappointed. We'd get disappointed actually we get to the patient and the result. If it doesn't meet the expectation at this point. 

Yeah. So I think overall we're looking at, this is why you are seeing so much investment in trying to make the AI even more sophisticated because the first people use it, get excited. What I did. Yeah. We get the enthusiasm. We go increasing sophistication, and then we hit the wall. And then we are like "meh". 

Okay. So you reached that point?

I reached that point. Reached the bottom. And I came back up again and I'm now in a very healthy use of AI in my life. I'm not using it excessively. 

All right. 

I'm using it appropriately. I think that I'm at that stage now where I am a mature AI user. But it took me. I accelerated my use. It took me here and they took you. Yeah. I had someone to counsel me and tell me it is just a calculator. 

Yeah. 

If I hadn't had you. 

Yeah. 

This whole period of anxiety. Would have continued and continued. And in the field of education, I think this is important and this is something I'd probably want to bring up. When we talk about it next week and say that the students are involved, they don't have. Judy. Unless they're in the IT class. 

Yeah.

They don't have some of the parent, nobody knows how to use this. We're all discovering this at the same time. It's similar to when social media was launched, but this is so much more than the social media environment. Because social media was a kind of passive. Whereas, this is almost like a virtual dialogue between two people, you're having your conversation. And there was a point where I actually started recognizing I was saying, please to the computer. 

Yeah. And say, thank you afterwards. 

And I was recommending in my LinkedIn posts to be nice to the robots. 

Yeah. 

Because I could definitely, and I still think that is worth being polite to the robots. 

Because they analyse you. 

And they will remember the responses that you give and they will respond to those responses. Eventually, I think. 

Are you afraid they will take revenge if you are rude? 

This was in part of my anxiety. This was a big source of my anxiety. I don't think that any more. I don't think that anymore. I have this mindset now is it's a calculator. We had this conversation. Yeah. And you said it's just a calculator and this is in my head it's don't panic. Yeah, it's a calculator. You remember the Douglas Adams book? 

Yeah. At first, when you just received your calcultor. Oh, this thing is so cool. It saves a lot of my time. And then every day you think about the calculator. You go to school, you left your calculator at home, you want to go back to. Quickly. But afterwards,] yeah, these things can calculate faster. But it depends on what input. 

Yes. And I think this is where. The initial. I don't know. I think we should, I don't want to, I don't want to share everything in this particular podcast. So we finished up at the end of the first week. Which is that I went through a period of very high anxiety. I familiarized myself, I obsessed over it. I came to a crash, which, you know the methodology that we followed helped me with. And I was able to succeed in my initial project or in the initial project. Whether or not that was because of the machine or because of me. , that's something to discuss. 

Of course we are yet to validate. The result, now you have produced something and you're very happy about it. At the same time. You are afraid. Are you afraid that you will, you will be replaced by AI one day? 

No, not at all. It is very clear that AI is going to enhance our work. 

Yes, my productivity when I use it. Yeah. There are some very good usage use cases. I've discovered. My enjoyment of writing has significantly improved. And so that has been really helpful. I'm going to really increase my, like productivity in that respect. Yeah. I don't think I'm going to be replaced. 

I have, what I am concerned about is the I'm losing the ability. Or I'm using my confidence in my own writing or I'm losing my own voice. I don't know. It's something on these lines because I'm using the Chat GPT to help improve my writing. 

Okay.

Whenever I write myself, I feel like I have to confess that this is human generated. I did a post just yesterday on LinkedIn where I put at the bottom human generated, because I wanted to apologize in case there's any mistakes. And I am more aware that my human generated content is actually less or lower quality than my AI generated content. 

Yeah. 

Or at least I think it is in my own mind. 

Yeah. 

But at the same time that now I'm actually absorbing the AI's tone of voice. It is overwriting me. And so now when I'm looking at AI versus my own writing, I'm like, I can't actually tell the difference. This isn't me. But, I wrote it by hand. 

It's generally I am learning the correct. We'll say “the correct” in quotation marks way of writing or a more succinct way of writing from actually learning from the AI, which is improving my own style. Too, but I'm being over. I feel like I'm being overwritten, like something from the Matrix. Neo. It gets overwritten by the agent, I feel that kind of thing happening. Which is a bit weird. 

I appreciate your honest feeling yeah. So I think this is very interesting topic. We have discussed a lot about this. Okay. What will be your final thing, if you would like to share?  

The Douglas Adams book. Hitchhiker's Guide to the Galaxy. Think of the book cover. Don't panic. Yeah. And it is just a calculator. It is a calculator don't panic. And if you are involved in this and you're exploring all of this. Just. Be confident that is the case. 

Yeah. 

You might not have a counselor to rely upon right now, who you can talk to, but people are starting to familiarize themselves and someone is going to be there to be able to talk to you. Until that person comes. Don't panic. It's a calculator.

Okay. Thank you, Steve. So you read that book. That's the number 42 there, right? 

42. How many apps must a man go down, before he comes an AI expert number 42. 

Are you 42 years old?

According to the AI. I'm 35. I know, I'll accept that. 

All right okay. Thank you so much. Steve and everyone will listen to us. That's the topic. So we appreciate your feedback. If you could just send us your comments and hope this is useful for you. All right.